Lenny's Podcast: Product | Career | Growth - Why AI evals are the hottest new skill for product builders | Hamel Husain & Shreya Shankar (creators of the #1 eval course)

Episode Date: September 25, 2025

Hamel Husain and Shreya Shankar teach the world’s most popular course on AI evals and have trained over 2,000 PMs and engineers (including many teams at OpenAI and Anthropic). In this conversation, they demystify the process of developing effective evals, walk through real examples, and share practical techniques that’ll help you improve your AI product.What you’ll learn:1. WTF evals are2. Why they’ve become the most important new skill for AI product builders3. A step-by-step walkthrough of how to create an effective eval4. A deep dive into error analysis, open coding, and axial coding5. Code-based evals vs. LLM-as-judge6. The most common pitfalls and how to avoid them7. Practical tips for implementing evals with minimal time investment (30 minutes per week after initial setup)8. Insight into the debate between “vibes” and systematic evals—Brought to you by:Fin—The #1 AI agent for customer serviceDscout—The UX platform to capture insights at every stage: from ideation to productionMercury—The art of simplified finances—Where to find Shreya Shankar• X: https://x.com/sh_reya• LinkedIn: https://www.linkedin.com/in/shrshnk/• Website: https://www.sh-reya.com/• Maven course: https://bit.ly/4myp27m—Where to find Hamel Husain• X: https://x.com/HamelHusain• LinkedIn: https://www.linkedin.com/in/hamelhusain/• Website: https://hamel.dev/• Maven course: https://bit.ly/4myp27m—In this episode, we cover:(00:00) Introduction to Hamel and Shreya(04:57) What are evals?(09:56) Demo: Examining real traces from a property management AI assistant(16:51) Writing notes on errors(23:54) Why LLMs can’t replace humans in the initial error analysis(25:16) The concept of a “benevolent dictator” in the eval process(28:07) Theoretical saturation: when to stop(31:39) Using axial codes to help categorize and synthesize error notes(44:39) The results(46:06) Building an LLM-as-judge to evaluate specific failure modes(48:31) The difference between code-based evals and LLM-as-judge(52:10) Example: LLM-as-judge(54:45) Testing your LLM judge against human judgment(01:00:51) Why evals are the new PRDs for AI products(01:05:09) How many evals you actually need(01:07:41) What comes after evals(01:09:57) The great evals debate(1:15:15) Why dogfooding isn’t enough for most AI products(01:18:23) OpenAI’s Statsig acquisition(1:23:02) The Claude Code controversy and the importance of context(01:24:13) Common misconceptions around evals(1:22:28) Tips and tricks for implementing evals effectively(1:30:37) The time investment(1:33:38) Overview of their comprehensive evals course(1:37:57) Lightning round and final thoughts—LLM Log Open Codes Analysis Prompt:Please analyze the following CSV file. There is a metadata field which has an nested field called z_note that contains open codes for analysis of LLM logs that we are conducting. Please extract all of the different open codes. From the _note field, propose 5-6 categories that we can create axial codes from.—Referenced:• Building eval systems that improve your AI product: https://www.lennysnewsletter.com/p/building-eval-systems-that-improve• Mercor: https://mercor.com/• Brendan Foody on LinkedIn: https://www.linkedin.com/in/brendan-foody-2995ab10b• Nurture Boss: https://nurtureboss.io/• Braintrust: https://www.braintrust.dev/• Andrew Ng on X: https://x.com/andrewyng• Carrying Out Error Analysis: https://www.youtube.com/watch?v=JoAxZsdw_3w• Julius AI: https://julius.ai/• Brendan Foody on X—“evals are the new PRDs”: https://x.com/BrendanFoody/status/1939764763485171948• Who Validates the Validators? Aligning LLM-Assisted Evaluation of LLM Outputs with Human Preferences: https://dl.acm.org/doi/abs/10.1145/3654777.3676450• Lenny’s post on X about evals: https://x.com/lennysan/status/1909636749103599729• Statsig: https://statsig.com/• Claude Code: https://www.anthropic.com/claude-code• Cursor: https://cursor.com/• Occam’s razor: https://en.wikipedia.org/wiki/Occam%27s_razor• Frozen: https://www.imdb.com/title/tt2294629/• The Wire on HBO: https://en.wikipedia.org/wiki/The_Wire—Recommended books:• Pachinko: https://www.amazon.com/Pachinko-National-Book-Award-Finalist/dp/1455563935• Apple in China: The Capture of the World’s Greatest Company: https://www.amazon.com/Apple-China-Capture-Greatest-Company/dp/1668053373/• Machine Learning: https://www.amazon.com/Machine-Learning-Tom-M-Mitchell/dp/1259096955• Artificial Intelligence: A Modern Approach: https://www.amazon.com/Artificial-Intelligence-Modern-Approach-Global/dp/1292401133/Production and marketing by https://penname.co/. For inquiries about sponsoring the podcast, email [e-mail].—Lenny may be an investor in the companies discussed.My biggest takeaways from this conversation: To hear more, visit www.lennysnewsletter.com

To build great AI products, you need to be really good at building e-vals. It's the highest ROI activity you can engage in. This process is a lot of fun. Everyone that does this immediately gets addicted to it when you're building an AI application. You just learn a lot. What's cool about this is you don't need to do this many, many times. For most products, you do this process once and then you build on it. The goal is not to do e-vows perfectly.

It's to actionably improve your product. I did not realize how much controversy and drama there is around e-vals. There's a lot of people of very strong opinions. People have been burned by e-vowls in the world. past. People have done e-vals badly, and then they didn't trust it anymore, and then they're like, oh, I'm anti-evils. What are a couple of the most common misconceptions people have with evils? The top one is, we live in the age of AI. Can't the AI just eval it? But it doesn't work. A term that you used in your posts that I love is this idea of a benevolent dictator. When you're doing

this open coding, a lot of teams get bogged down in having a committee do this. For a lot of situations, that's wholly unnecessary. You don't want to make this process so expensive that you can't do it. You can appoint one person whose taste that you trust. It should be the person with domain expertise. Oftentimes it is the product manager. Today my guests are Hamil Hussein and Shreya Shankar. One of the most trending topics on this podcast over the past year has been the rise of e-vals. Both the chief product officers of Anthropic and Open AI shared that e-Vals are becoming the most important new skill for product builders. And since then, this has been a recurring theme across many of the top AI builders I've had on.

Two years ago, I had never heard the term e-vals. Now it's coming up constantly. When was the last time that a new skill emerged that product builders had to get good at to be successful? Hamel and Shreya have played a major role in shifting e-vals from being an obscure, mysterious subject to one of the most necessary skills for AI product builders. They teach the definitive online course on e-vals,

which happens to be the number one course on Maven. They've now taught over 2,000 PMs and engineers across 500 companies, including large swaths of the Open AI Ananthropic teams, along with every other major AI lab. In this conversation, we do a lot of show versus tell. We walk through the process of developing ineffective e-val, explain what the heck evals are and what they look like, address many of the major misconceptions with e-vals, give you the first few steps you can take to start building evils for your product, and also share just a ton of best practices that Hamill and Trey have developed over the past few years. This episode,

This episode is the deepest yet most understandable primer you'll find on the world at e-vals. And honestly, it got me excited to write e-vals, even though I have nothing to write e-vals for. I think you'll feel the same way as you watch this. If this conversation gets you excited, definitely check out Hamill and Shreya's course on Maven. We'll link to it in the show notes. If you use the code Lenny's List when you purchase the course, you'll get 35% off the price of the course. With that, I bring you Hamel Hussein and Shreya Shankar. This episode is brought to you by Finn, the number one AI agent for customer service.

If your customer support tickets are piling up, then you need Finn. Finn is the highest performing AI agent on the market with a 65% average resolution rate. Finn resolves even the most complex customer queries. No other AI agent performs better. In head-to-head bakeoffs with competitors, Finn wins every time. Yes, switching to a new tool can be scary, but Finn works on any help desk with no migration needed, which means you don't have to overhaul your current system or deal with delays in service for your customers.

And Finn is trusted by over 5,000 customer service leaders and top AI companies like Anthropic and Synthesia. And because Finn is powered by the Finns AI engine, which is a continuously improving system that allows you to analyze, train, test, and deploy with ease, Finn can continuously improve your results too. So if you're ready to transform your customer service and scale your support, give Finn a try

for only 99 cents per resolution. Plus, Finn comes with a 90-density. day money back guarantee. Find out how Finn can work for your team at f-in.a-i-slash Lenny. That's fin.com. This episode is brought to you by Descout. Design teams today are expected to move fast, but also to get it right. That's where DeScout comes in. DeScout is the all-in-one research platform built for modern product and design teams. Whether you're running usability tests, interviews, surveys, or in the wild fieldwork, D-Skout makes it easy to connect with real users and get real insights fast. You can even test your Figma prototypes directly inside the

platform. No juggling tools, no chasing ghost participants. And with the industry's most trusted panel plus AI-powered analysis, your team gets clarity and confidence to build better without slowing down. So if you're ready to streamline your research, speed of decisions, and design with impact, heads at D-Skout.com to learn more. That's dsc-o-u-t.com. The answers you need, to move confidently. Pamel and Shreya, thank you so much for being here. And welcome to the podcast. Thank you for having us.

Yeah, super excited. I'm even more excited. Okay, so a couple years ago, I had never heard the term e-vals. Now it's one of the most trending topics on my podcast, essentially that to build great AI products, you need to be really good at building e-vals. Also, it turns out some of the fastest growing companies in the world are basically building and selling and creating evals for a,

AI Labs. I just had the CF Mercor on the podcast. So there's something really big happening here. I want to use this conversation to basically help people understand the space deeply. But let's start with the basics. Just what the heck are evils for folks that have no idea what we're talking about. Give us just a quick understanding of what an eval is. And let's start with Hamel. Sure. Evils is a way to systematically measure and improve an AI application. And it really doesn't have to be scary or unapproachable at all. It really is at its core data analytics on your LLM application in a systematic way of looking at that data and where necessary, creating metrics around things so you can measure what's happening and then so you can iterate and do experiments

and improve. So that's a really good broad way of thinking about it. If you go one level deeper, to give people a even more concrete way of imagining and visualizing what we're talking about it. Even if you have an example to show it would be even better, what's an even deeper way of understanding what an eval is? Let's say you have a real estate assistant application and it's not working the way you want. It's not writing emails to customers the way you want or it's not calling the right tools or any number of errors. And before evals, you would be left with guessing. You would maybe fix a prompt and hope that you're not breaking anything else with that prompt. And you might rely on vibe checks, which is totally fine.

And vibe checks are good. And you should do vibe checks initially. But it can become very unmanageable, very fast. Because as your application grows, it's really hard to rely on vibe checks. You just feel lost. And so evals help you create metrics that you can use to measure how your application is doing and kind of give you a way to improve your application with confidence. You have a feedback signal in which to iterate against.

So just to make it very real, so imagining this real estate agent, maybe they're helping you book a listing or go see an open house. The idea here is you have this agent talking to people, It's answering questions, pointing them to things as a builder of that agent. How do you know if it's giving them good advice, good answers? Is it telling them things that are completely wrong? So the idea of e-vals essentially is to build a set of tests that tell you is how often are, is this agent doing something wrong that you don't want it to do?

And there's a bunch of ways wrong. You could define wrong. It could be just making up stuff. It could be just answering in a really strange way. The way I think about e-vall is and tell me if this is wrong, just simply is like unit tests. for code and the main, you're smiling, you're like, no, you idiot. No, that's not what I was thinking. Okay, okay, tell me. Tell me, how does that feel as a metaphor?

So, okay, I like what you said first, which is we had a very broad definition. Evales is a big spectrum of ways to measure application quality. Now, unit tests are one way of doing this. Maybe there are some non-negotiable functionalities that you want your AI assistant to have, and unit tests are going to be able to check that. Now, maybe you also, because these AI assistants are doing such open-ended tasks, you kind of also want to measure how good are they at very vague or ambiguous things, like responding to new types of user requests,

or figuring out if there's new distributions of data, like new users are coming and using your real estate agent that you didn't even know would use your product, and then all of a sudden you think, like, oh, there's a different way you want to kind of accommodate this new group of people. So evals could also be, you know, a way of looking at your data regularly to find these new cohorts of people. Evals could also be like metrics that, you know, you just want to track over time. Like you want to track people saying yes, thumbs up. I liked your message. You want a very, very basic things that are not necessarily AI related, but can go back into

this flywheel of improving your product. So I would say, on the overall, right, unit tests are a very small part of that very big puzzle. Awesome. You guys actually brought an example of an eval, just to show us exactly what we're talking about. We're talking in these big ideas. So how about it's pull one up and show people here's what an eval is. Yeah, let me just set the stage for it a little bit. So to echo what Shrea said, it's really important that we don't think of evals as just tests. There's a common trap that a lot of people fall into because they jump straight to the test, like, let me write some tests.

And usually that's not what you want to do. You should start with some kind of data analysis to ground what you should even test. And that's a little bit different than software engineering where you have a lot more expectations of how the system is going to work. With LLMs, it's a lot more surface area. It's very stochastic. So we kind of have a different flavor here. And so the example I'm going to show you today, it's actually a real estate example.

It's a different kind of real estate example. It's from a company called NurtureBoss. I can share my screen to show you their website. Just to help you understand this use case a little bit. So let me share my screen. So this is a company that I worked with called NurtureBoss, and it is a AI assistant for property managers who are managing apartments. and it helps with various tasks such as inbound leads, customer service, booking appointments, so on and so forth.

It's like all the different sort of operations you might be doing as a property manager. It helps you with that. And so, you know, you can see kind of what they do. It's a very good example because it has a lot of the complexities of a modern AI application. So there's lots of different channels that you can. interact through the AI with, like chat, text, voice, but also there's tool calls, lots of tool calls for like booking appointments, getting information about availability, so on and so forth. There's also rag retrieval, getting information about customers and properties and things

like that. So it's pretty fully fleshed in terms of an AI application. And so they have been really generous with me in allowing me to use their data as a teaching example. And so we have anonymized it, but what I'm going to walk through today is, okay, let's create, let's do the first part of how we would start to build evals for NurtureBoss. Like, why would we even want to do that? So let's go through the very beginning stage, what we call error analysis, which is, let's look at the data of their application and first start with what's going wrong. So I'm going to jump to that next. And I'm going to open an observability tool. And you can use whatever you want here.

I just happen to have this data loaded in a tool called brain trust. But you can load it in anything. You know, it's not, we don't have a favorite tool or anything in the blog post that we wrote with you, we had the same example, but in Phoenix, Arise. And I think Amman on your blog post use Phoenix arise as well. And there's also Langsmite. So these are kind of like different tools that you can use. So what you see here on the screen, this is logs from the application. And let me just show you how it looks. So what you see here is, and let me make it full screen, so this is one particular interaction that a customer had with the NurtureBoss application. And what it is, it's a detailed log of everything that happened. So it's called a trace. And it's just an engineering

term for logs of a sequence of events. The concept of a trace has been around for really long time, but it's especially really important when it comes to AI applications. So we have all the different components and pieces and information that the AI needs to do its job, and we are logged all of it. And we're looking at a view of that. And so you see here a system prompt. The system prompt says, you are an AI assistant working as a leasing team member at retreat at Acme Apartments. Remember, I said, this is anonymized. So that's why the name is Acmey Apartments. departments. Your primary role is to respond to text messages from both residents and prospective, both current residents and prospective residents. Your goal is to provide accurate,

helpful information, yada, yada, yada. And then there's a lot of detail around guidelines of how we want this thing to behave. Is this their actual system prompt, by the way, for this company? It is. Yes, it is. It's a real system prompt. That's amazing because that's really, it's rare you see actual company product system prompt. That's like they're crowned jewels a lot of time. So this is actually very cool on its own. Yeah, yeah, it's really cool. And, you know, you see all these different sort of features that they want to are different use cases. So things about tour scheduling, handling applications, guidance on how to talk to different personas, so on and so forth. And you can see the user just kind of jumps in here. It says, ask, okay, do you have a one

bedroom with study available, I saw it on virtual tours. And then you can see that the LLM calls some tools. It calls this get individuals information tool and it pulls back that person's information. And then it gets the community's availability. So it's, you know, it's querying a database with the availability for that apartment complex. And then finally, the A.A. AI responds, hey, we have several one-bedroom apartments available, but none specifically listed with a study. Here are a few options. Then it says, can you let me know when one with a study is available? Then it says, I currently don't have specific information on the availability of a one-bedroom apartment. User says, thank you. And the AI says, you're welcome. If you have any more

questions, feel free to reach out. Now, this is an example of a trace. And this is, we're looking at one specific data point. And so one thing that's really important to do when you're doing data analysis of your LLM application is to look at data. Now, you might wonder, there's a lot of these logs. It's kind of messy. There's a lot of things going on here. How in the hell are you supposed to look at this data? Do you want to just drown in this data? How do you even analyze this data? So it turns out there is a way to do it that is completely manageable.

And it's not something that we invented. It's been around in machine learning and data science for a really long time. And it's called error analysis. And what you do is the first step in conquering data like this is just to write notes. Okay? So you've got to put your product hat on, which is why we're talking to you, because product people have to be in the room. And they have to be involved in sort of doing this. You know, usually a developer is not suited to do this, especially it's not a coding application.

And I'm just to mirror back why I think you're saying that is because this is the user experience of your product. People talking to this agent is the entire product essentially. And so it makes sense for the product person to be involved, super involved in this. Yeah. So let's reflect on this conversation. Okay, a user asked about availability. The AI said, oh, we don't really have that. Have a nice day. Now, for a product that is helping you with lead management, is that good?

Like, do you feel like this is the way we want it to go? Not ideal. Yes, not ideal. And I'm glad you said that. A lot of people would say, oh, it's great. Like, the AI did the right thing. It said we don't, it looked, said we didn't have available, and it's not available. But with your product had on, you know that's not correct. And so what you would do is you would just write a quick note here. You would say, okay, you know, you might pop in here. Let me just, and you can write a note. So every observability application has ability to write notes. And you wouldn't try to figure out if something is wrong in this application. In this case, it's kind of not doing the right thing. But you just write a quick note, should have handed off to a human. And as we watch this happening, it's like you mentioned this and you'll explain more. You're doing this.

This feels very manual and unsc scalable. But as you said, this is just one step of the process and there's a system to this. And it's just the first part. and you don't have to do it for all of your data. You can sample your data and just take a look. And it's surprising how much you learn when you do this. Everyone that does this immediately gets addicted to it and they say this is the greatest thing that you can do

when you're building an AI application. You just learn a lot. You're like, hmm, this is not how I want it to work. Okay. And so that's just an example. So you write this note. And then we can go on to the next trace. So this is the next trace.

I just pushed a hot key on my keyboard. Let me go back to looking at it. And these tools make it easy to go through a bunch and add these notes quickly. Yes. And so this is another one. Similar system prompt. We don't need to go through all of it again.

We'll just jump right into the user question. Okay, I've been texting you all day. Maybe it's funny. And the user says, please, okay, yeah, this one is, this one is just like an error in the application where, you know, this is a text message application. And so, you know, it's a tech, the, sorry, the channel through which the customer is communicating is through text message and you're just getting like really garbled.

And you can see here that it kind of doesn't make sense. you know, like the words are being cut off, like in the meantime. And then the system doesn't know how to respond. Because you know how people text message. They like write short phrases. They, you know, split their sentence across four or five different turns. So in this case... What do you do with something like that?

Yeah. So this is a different kind of error. This is more of, hey, we're not handling this interaction correctly. It's more of a technical problem. Rather than, hey, the AI is not doing exactly what we're. we want. So we would write that down to. Like it's amazing you're catching that too here. Otherwise, you have no idea this is happening. Yeah, you might not know this is happening, right? And so you would just say, okay, you would write a note. Like, oh, conversation flow is janky because of text

message. And I like, yeah, I like that. I like that you using the more janky. It shows you just how informal this can be at this stage. Yeah, it's supposed to be chill. Like, just don't overthink it. And there's a way to do this. So the question always comes up. How do you do this? Do you look at, do you try to find all the different problems in this trace? What do you write a note about? And the answer is, just write down the first thing that you see that's wrong, the most upstream error.

Don't worry about all the errors. Just capture the first thing that you see that's wrong and stop and move on. and you can get really good at this. The first two or three can be very painful, but, you know, it doesn't, we can, you know, do a bunch of them really fast. So here's another one. And let's skip the system prompt again.

And the user asks, hey, I'm looking for a two to three bedroom with either one or two bats. Do you provide virtual tours? And a bunch of tools are called. And it says, Hi, Sarah. Currently, we have three bedroom, two and a half bathroom apartment available for $2,175.

Unfortunately, we don't have any two bedroom options at the moment. We do offer virtual tours. You can schedule a tour, blah, blah, blah. It just so happens that there's no virtual tour. Right? So, you know, it is hallucinating something that doesn't exist. And you would, you kind of have to bring your context as an engineer or even, you know, a product content and say,

this is kind of weird, like, you know, we shouldn't be telling a person about virtual tour when it's not offered. So you would say, okay, you know, offered virtual tour, and you just, you know, you just write the note. So you can see there's a diversity of different kinds of errors that we're seeing. And we're actually learning a lot about your application in a very short amount of time. One common question that we get from people at this stage is, is, okay, I understand what's going on. Can I ask an LLM to do this process for me? Great question.

And I loved Hamel's most recent example because what we usually find when we try to ask an LLM to do this error analysis is it just says the trace looks good. Because it doesn't have the context needed to understand whether something might be, you know, bad product smell or, you know, not, for example, the hallucination about scheduling the tour, right?

I can guarantee you, I would bet money on this. If I put that into chat, GPT and asked, is there an error? It would say, no, did a great job. But Hamel had the context of knowing, oh, we don't actually have this virtual tour functionality. Right. So I think in these cases, it's so important to make sure you are manually doing this yourself. And we'll talk a little bit more about when to use LLMs in the process later. But like, number one pitfall right here is people are like, let me automate this with an LLB.

Do you think they'll get to a place where we're in a lot? agent can do this? Oh, no, no, no, no. Sorry, there are parts of error analysis that an LLM is suited for, which we can talk about later in this podcast. But right now in this stage of free form note-taking is not the place for an LLM. And this is something you call open coding. Yes, absolutely. Another term that used in your posts that I love and that fits into the step is this idea of a benevolent dictator. Maybe just talk about what that is and maybe Sharia cover that. So Hamill actually came up with this term. Okay, maybe Hamill cover the actual.

No problem. And we'll actually show the LM automation in this example. Because we're going to take this example. We're going to go all the way through. Amazing. And so benevolent dictator is just a catchy term for the fact that when you're doing this open coding, a lot of teams get bogged down in having a committee do this. and for a lot of situations, that's wholly unnecessary.

Like, you know, people get really uncomfortable with, okay, you know, we want everybody on board, we want everybody involved, so on and so forth. You need to cut through the noise. In a lot of organizations, if you look really deeply, especially small and medium-sized companies, there's really, like, you can appoint one person whose taste that you trust.

and you can do this with a small number of people and often one person. And that's, it's really important to make this tractable. You don't want to make this process so expensive that you can't do it. You're going to lose out. So that's the idea behind benevolent dictator is, hey, you need to simplify this across as many dimensions as you can. Another thing that we'll talk about later is when you goes to building an LLM as a judge, you need a binary score. You don't want to think about, is this like a one, two, three, four, five, like assigned a score to it. You can't, that's going to slow it down.

Just to make sure this benevolent dictator point is really clear. Basically, this is the person that does this note taking. And ideally, they're the expert on the stuff. So if it's law stuff, maybe there's like a legal person that owns this. It could be a product manager. Give us advice on who this person should be. Yeah, it should be the person with domain expertise. So in this case, you know, it would be.

the person who understands the business of apartment leasing and has context to understand if this makes sense it's always a domain expert like you said okay for legal it would be a law person for mental health it would be the mental health expert whether that's like a psychiatrist or you know someone else cool um oftentimes it is the product manager cool so the advice here pick that person may not feel so super fair that they're the one in charge and they're the dictator but they're benevolent it's going to be okay.

Yeah, it's going to be okay. You're just trying to, it's not perfection. You're just trying to make progress and get signal quickly. So you have an idea of what to work on because it can become infinitely expensive if you're not careful. Yeah. Okay, cool. Let's go back to your examples. Yeah, no problem.

So this is another example where we have someone saying, okay, do you have any specials? and the assistant or the AI responds, hey, we have a 5% military discount. User response, can you, and the switch is a subject, can you tell me how many floors there are? Do you have any one bedrooms available, or one bedrooms on the first floor?

And the AI responds, yeah, okay, we have several one bedroom apartments available. And then the user wants to confirm any of those on the first floor, and how much are the one bedrooms? And then also is a current resident. So they're also asking, I need a maintenance request. This is actually pretty, like, you could see the messiness of the real world in here. And the assistant just calls a tool that says transfer call, but it doesn't say anything.

It just abruptly does transfer call. So it's pretty jank, I would say. Like, it's just not, you know. Another kind of jank. A different kind of jank. So you don't want to, when you write the open note, you don't want to say jank. Because what we want to do is we want to understand what and when we look at the notes later on, we'll understand like what happened. So you just sort of say, you know, did not confirm call transfer with user.

It doesn't have to be perfect. You just have to have a general idea of what's going on. Cool. So, okay. So let's say we do, and we, Treya and I, we recommend doing at least 100 of these. The question is always like, how many of this do you do? And so there's not a magic number. We say a 100 is because we know that as soon as you start doing this,

once you do 20 of these, you will automatically find it so useful that you will continue doing it. So we just say 100 to mentally unblock you so it's not intimidating. Like, don't worry, you're only going to do 100. And there is a term for that of, so the right answer is keep looking at traces until you feel like you're not learning anything new. Maybe Shreya should talk about. Yeah, so there's actually a term in data analysis and qualitative analysis called theoretical saturation. So what this means is when you do all of these processes of looking at your data, when do you stop? it's when you are theoretically saturating, or you're not uncovering any new types of notes,

new types of concepts, or nothing that will like materially change the next part of your process. And this kind of takes a little bit of intuition to develop. So typically people don't really know when they've reached theoretical saturation yet. That's totally fine. When you do two or three examples or rounds of this, like you will develop the intuition. A lot of people realize like, oh, okay, like I only need to do 40. I only need to do 60. Actually, I only need to do like 15.

I don't know. Like depends on the application and develops like how, depends on how savvy you are with error analysis for sure. And your point about you probably want to, you're going to want to do a bunch. I imagine it's because you're just like, oh, I'm discovering all these problems. I got to see what else is going on here. Exactly. And promise at some point you're like not going to discover new types of problems. Yeah.

Awesome. So let's say you did 100 of these. with the next step. Yeah, okay, so you did 100 of these. Now you have all these notes. So this is where you can start using AI to help you. So the part where you looked at this data is important, like we discussed, you don't want to automate this part too much.

Humans will still have jobs. This is a takeaway here. That's great. Yes, just reviewing traces. At least there's one job left for now. Yeah, exactly. And so, okay, you have all these notes. Now, to turn this into something useful, you can do basic counting.

So basic counting is the most powerful analytical technique in data science. Because it's so simple and it's kind of undervalued in many cases. And so it's very approachable for people. And so the first thing you want to do is take these notes and you can categorize them with an LM. And so there's a lot of different ways to do that. right before this podcast, I took three different coding agents or, you know, AI tools and had it categorize these notes. So one is, okay, I uploaded it into a cloud project. I uploaded a CSV of these notes. And I just exported them directly from this interface.

There's a lot of different ways to do this, but I'm showing you the simple, stupid way, the most basic way. the most basic way of doing things. And so I dumped a CSV in here, and I said, please analyze the following CSV file. And I told it there's a metadata field that has a note in it. But what I said is I use the word open codes. And I said, hey, I have different open codes. And that's a term of art.

That's LMs know what open codes are and they know what axial codes are, because it is a concept that's been around for a really long time. So those words help me shortcut what I'm trying to do. That's awesome. And the end of the prompt is telling it to create axial codes. Yes, creating axial codes. So what it does is...

So maybe it's worth talking about what are axial codes or like, what's the point here? You have a mess of open codes, right? And you don't have 100 distinct problems. Actually, many of them are repeats, but because... because you phrased them differently, right? And you shouldn't have tried to create your taxonomy of failures as your open coding. You just want to get down what's wrong and then organize, okay, what's the most common failure mode?

So the purpose, axial code basically is just a failure mode. It's like the label or category. And what our goal is is to get to this clusters of failure modes and figure out what is the most prevalent. So then you can go and run and attack that problem. That is really helpful. Basically, just synthesizing all these into categories and themes. Super cool.

And we'll include this prompt in our show notes for folks so they don't have to sit there and screenshot it and try to type it up themselves. Yeah, great idea. And so Claude, you know, went ahead and analyzed the CSV file, decided how to parse it, blah, blah, blah. We don't need to worry about all that stuff. But it came up with a bunch of axial codes. Basically, axial codes are categories, like Shrey has said. So one is, okay, capability limitations, misrepresentation, process of protocol violations, human handoff issues, communication quality. It created these categories. Now, do I like all the categories?

Not really. I like some of them. It's a good first, like, stab at it. I would probably rename it a little bit, because some of them are a bit too generic. Like, what is capability limitations? I said a little bit too broad. It's not actionable. I want to get like a little bit more actionable with it so that if I do decide as a problem, I know what to do with it. But we'll discuss that in a little bit. So you can do this like with anything. And this is the dumbest way to do it. But dumb sometimes is a good way to get started. And this is what LMs are really good at, taking a bunch of information and synthesizing. Absolutely. Sythesizing for us to make sense of, right? Note that, you know, it's not telling us, it's not automatically proposing fixes or anything. That's our job. But, you know, now we can wait through

this mess of open codes a lot easier. Another thing that's interesting here in this prompt to generate the axial codes is you can be very detailed if you want, right? You can say, I want each axial code to actually be, you know, some actionable failure mode. And maybe the LLM will understand that and propose it. Or I want you to group these open codes by, you know, what stage of the user story that it's in. So this is where you can, you know, be creative.

or do what's best for you as a product manager, engineer working on this, and that will help you do the improvement later. So there's no definitive prompt of here's the one way to do it. You're saying there's, you can iterate, see what works for you. Absolutely. It's interesting the tools don't want to do this, or do they try and they just don't do a great job? No, I don't think they do it. We've been screaming from the rooftops.

Please, please do this. I do think it's a little bit hard, right? Like part of this whole experience with the Eval's course, Hamel and I are teaching, or like a lot of people don't actually know this. So maybe it's that people don't know this and they don't know how to build tools for it. And hopefully we can demystify some of this magic. And just to double click on this point,

like this is not a thing everyone does or knows. This is something you too developed based on your experience doing data analysis and data science and other companies. Well, I want to caveat. We didn't invent error analysis. We don't actually want to invent things. That's a bad, that's bad signal.

If somebody is coming to you with a way to do something that's like entirely new and not grounded in hundreds of years of theory and literature, then you should, I don't know, be a little bit wary of that. But what we tried to do was distill, okay, what are the new tools and techniques that you need to make sense of the LLM error analysis? And then we created a curriculum or structured way of doing this. So this is all very tailored to LLMs. but the terms open coding, axial coding, are grounded in social science. Amazing. Okay. Like, what's funny about you do guys doing this is I just want to go do this somewhere.

I don't have any product to do this on, but it's just like, oh, this would be so fun. Just sit there and find all the problems I'm running into it and categorize them and then try to fix them. I love that. Hamel pulled up a video. What do you got going on here? Yeah, so I pulled up a video just to drive home Shreya's point.

like we are not inventing anything. So what you see on the screen here is Andrew Eng, one of the famous machine learning researchers in the world who have taught a lot of people, frankly, machine learning. And you can see this is an eight-year-old video. And he's talking about error analysis. And so this is a technique that's been used to analyze stochastic systems for ages.

And it's something that if you're just using the same, machine learning ideas and principles is bringing them into here. Because again, these are stochastic systems. Awesome. One thing we're working on getting Andrew in the podcast, we're chatting, so that'll be really fun. Two, I love that my other, my podcast episode just came out today is in your feed there, and it's standing out really well in that feed. So I'm really happy about that thumbnail. Very nice. Yeah, the recommendation algorithm is quite good. Yes, here we go. You hope you click on that. Don't screw my algorithm. Okay, cool. So we've done some synthesis. I know we're not going to go through the step. This is like, you have a whole course that takes many days to learn this whole process.

What else do you want to share about how to go about this process? Okay, so you can do this through anything and, you know, I've used the same thing works just fine in chat GPT, the same exact prompt. You can see it made axial codes. I really like using Julius AI. This one of my favorite tools. Julius is a kind of its third-party tool, but it uses notebooks. I personally like Jupiter notebooks a lot. And so it's more of a data science thing, but a lot of product managers that are kind of learning notebooks nowadays, and it's kind of cool. It's like a fun playground where you can write code and look at data. We don't have to go deeply into that.

I just wanted to mention you can use a lot, you know, AI is really good at this. So let's go to the fun part. Here we go. So now we have all the axi, we have these axial codes. So the first thing I like to do, I have these open codes, right? and I have the axial codes that, let's say, you know, the, like, that we assigned from the Cloud Project or the chat GPT. And so what I do is I collect them first, and I take a look, like, does these axial codes

that make sense? And I look at the correspondence between the different axial codes and the open codes, and I go through an exercise and I say, hmm, do I like these codes? Like, can I make them better? can I refine them, can make them more specific. Instead of being generic, I make the very specific and actionable. So you see the ones that I came up with here are tour scheduling, rescheduling issues, human handoff or transfer issue, formatting error with an output, conversational flow.

We saw the conversational flow issue with the text messages, making follow promises not kept. And so basically what I can do, what you can do now, is like you have these axial codes. And so I just collect them into a list. So this is an Excel formula. Just collects these codes into a list. And now we have a comma, separated list of these codes. And then what you can simply do is you could take your notes that you have,

those open codes. And you can tell an AI, and this is using Gemini and AI, just for simplicity, this is like the, you know, again, we're trying to keep it simple. categorize the following note into one of the following categories. By the way, for folks watching, there's like,

I like all these different prompts and formulas you're sharing. This is like the Google Sheets AI AI prompt. Huge fan. Yeah. Mm. And so basically what you can do is you can then have,

you can categorize your traces into one of the buckets. And that's what we have here. We have categorized all those problems that we encountered into one of these things. And this is automatic, which is very exciting. I mean, the AI is doing it. So this also drives home the point that your open codes have to be detailed, right?

You can't just say janky because if the AI is reading janky, it's not going to be able to categorize it. Even a human wouldn't, right? It would have to go and remember why you said janky. So it's important to be, you know, somewhat detailed in your open code. Okay, so avoid the word janky is a good rule of thumb. Or have it with like another word. I was being funny.

Yeah, okay. What are some of those other words just that people often use that you think are not good? I don't think it's specific words. I think it's just people are not detailed enough in the open codes, so it's hard to do the categorization.

Great. And by the way, the reason you have to map them back is because, say, Collater of Japiti gave you suggestions and you change them and iterated on them. So you can't just go back and say, cool, whatever in the huge bucket. Yeah, yeah.

Great. That's a really good question, actually. It's good to iterate and think about it a little bit. Like, do I like these open codes? Do these actually make sense to me? Just like anything that AI does, it's really good to kind of put yourself in the middle. Just a little bit.

Still space for us. Great. One of the things that I like to do in this step, if I'm trying to use AI to do this labeling, is also have a new category called none of the above. So an AI can actually say none of the above in the axial code. and that informs me, okay, my axial codes are not complete. Like, let's go look at those open codes.

Let's figure out what some new categories are or figure out how to reward my other axial codes. Awesome. And what's cool about this is you don't need to do this many, many times. Like, for most products, you do this process once and then you build on it, I imagine, and you just tweak it over time. Absolutely. And it gets so fast. Like, people do this like once a week and you can do all of this in like 30 minutes. and like suddenly your product is like so much better than if you were never aware of any of these problems.

Yeah, it's absurd to feel like you don't, you wouldn't know this is happening. Like watching this happening, I'm like, how could you not do this to your product? A lot of people have no idea. Most people, yeah. We'll talk about that. There's a whole debate around this stuff that we want to talk about. Okay, cool. So you have this, you have this sheet.

I'll come next. Okay, so here's the big unveil. This is the magic moment right now. So we have all these codes that, you know, we applied, the ones that we like on our traces. Now you can do the to-da, you can count them. So here's a pivot table. And we just can do pivot table on those. And we can count how many times those different things occurred.

So what do we find? Find on this, on these like traces that we categorize, we found 17 conversational flow issues. And I really like pivot tables because you can do cool things. You can double click on these. You can say, oh, okay, let me take a look at those. But that's going into an aside about pivot tables, how cool they are. But, you know, now we have just a nice rough cut of what are our problems. And now we have gone from chaos to some kind of thinking around, oh, you know what?

these are my biggest problems. I need to fix conversational issues. You know, maybe these human handoff issues. But not necessarily the count is the most important thing. You know, that might be something that's just really bad and you want to fix that. But, okay, now you have some way of looking at your problem. And now you can think about whether you need evals for some of these. So, you know, with the, you know, there might be some of the, you know, there might be some of the,

of these things that might be just dumb engineering errors that you don't need to write an eval for because it's very obvious on how to fix them. Maybe the formatting error with output, maybe you just forgot to tell the LLM how you wanted to be formatted. And like, you didn't even say that in the prompt. So like, just go ahead and fix the prompt, maybe, you know? And we can decide, like, okay, do you want to write an e-val for that? You might still want to write an e-vail for that because you might be able to test that with just code. You could just test the string.

Does it have the right formatting, potentially, without running an LLM? So there's a cost-benefit trade-off to e-vowes. You don't want to get carried away with it, but you want to start, you want to usually ground yourself in your actual errors. You don't want to skip this step. And so the reason I'm kind of spending so much time on this is like, this is where people get lost.

They go straight into e-vals. Like, let me just write some tests. And that is where things go off the rails. So let's, let's say we want to tackle one of these things. So, for example, let's say we want to tackle this human handoff issue. And we're like, hmm, I'm not really sure. sure how to fix this. Like, that's a kind of subjective sort of judgment call on, you know, should we be handing off to human? And I don't know immediately how to fix it. It's not super

obvious per se. Yeah, I can like change my prompt, but I'm not like sure. I'm not 100% sure. Well, that might be sort of an interesting thing for an LLM as a judge, for example. So there's different kinds of e-vals. One is code-based. which you should try to do if you can because they're cheaper. You don't have to, you know, LM as a judge is something, it's like a meta-eval. You have to eval that e-vow to make sure the L-LM that judging is doing the right thing, which we'll talk about a second.

So, okay, LM as a judge. That's one thing. Okay, how do you build an LM as a judge? Before we get into that, actually, just to make sure people know exactly what you're describing there, these two types of evals. You said it's code-based, one is an LLM as judge. Maybe Shreya just help us understand what code-based e-val even is.

It's just like essentially a unit test. Is that a simple way to think about it? Maybe eval is not the right term here, but think like automated evaluator. So when we find these failure modes, one of the things we want is like, okay, can we now, like, go check the prevalence of that failure mode in an automated way without me manually labeling and doing all the coding and the grouping and I want to run it on thousands and thousands of traces. I want to run it every week. that is okay, you should probably build an automated evaluator to check for that failure mode. Now, when we're saying code-based versus L-LM-based, we're saying, okay, so maybe I could write like a Python function or a piece of code to check

whether that failure mode is present in a trace or not. And that's possible to do for certain things like checking the output is JSON, or checking that it's marked down or checking that it's short. Like these are all things you can capture in code or you could approximately, capture and quote. When we're talking about LLM judge here, we're saying that this is a complex failure mode and we don't know how to evaluate in an automated way. So maybe we will try to use an LLM to evaluate this very, very narrow, specific failure mode of handoffs. So just to try to mirror back where you're describing, you want to test what your, say, agent or AI product is doing. You ask

it a question. It gets back with something. One way to test if it's giving you the right answer is if it's consistently doing the same thing that you could write a code to tell you this is true or false. For example, will it ever say there's a virtual tour? So you could ask it. Do you provide virtual tours? It says yes or no. And then you could write code to tell you if it's correct based on that specific answer. But if you're asking about something more complicated and it's not binary, you almost need like in a one world, you need a human to tell you, this is correct. The solution to avoid humans having to review all this

every time automatically is LLM's replacing human judgment and you call it LLM as judge. The LM as being the judge if this is correct or not. Absolutely. You nailed it. So people always think like, oh, this is at least as hard of my problem

of creating the original agent and it's not. Because you're asking the judge to do one thing. Evaluate one failure mode. So the scope of the problem is very small. And the output of this LLM judge is like pass or fail. So it is a very, very tightly scoped thing that LLM judges are very capable of doing very reliably. And the goal here is just to have a suite of tests that run before you ship to production that tell you things are going

the way you want them to, the way your agent is interacting is correct. The beautiful thing about LLM judges, you can use them in unit tests or CI, sure. But you could also use it on line for monitoring, right? Like, I can sample like a thousand traces every day run my LLM judge, real production traces, and see what the failure rate is there. This is not a unit test, right? But still, now we get like an extremely specific measure of application quality. Cool. That's a really great point because a lot of people just evals for being this like not real life thing. It's a thing that you test before it's actually in the real world and what's actually happening in the real world. You're saying you could actually, you should actually do exactly that.

test your real thing running in production. And it's like a daily, hourly sort of thing you could be running. Totally. Awesome. Okay. Hamel's got an example of an actual LM as judge eval here. So let's take a look. I love how Shreya really teed it up for me.

So thank you so much. So what we have is a LM as a judge prompt for this one specific failure. Like Shreya said, you would want to do one specific failure and you want to make it binary. because we want to simplify things. We don't want, hey, like, score this on a rating of one to five, like, how good is it? That's just mostly, in most cases, that's a weasel way of, like, not making a decision. Like, no, you need to make a decision. Is this good enough or not?

Yes or no. It can be painful to think about what that is, but you should absolutely do it. Otherwise, this thing becomes very untractable. And then when you report these metrics, no one knows what 3.2 versus 3.7 means. Yeah, we see this all the time also. And even with like expert curated content on the internet where it's like, oh, here's your LLM judge evaluator prompt. Here's a one to seven scale. And I always think I always text Hamill like, oh, no, like now we have to fight the misinformation again because we know somebody is going to try it out and then come back to us and say, oh, I have 4.2 average.

And we're going to be like, okay. It's wild how much drama there is in eval space. We're going to get to that. Oh, man. This episode is brought to you by Mercury. I've been banking with Mercury for years. And honestly, I can't imagine banking any other way at this point. I switched from Chase and holy molly, what a difference.

Sending wires, tracking spend, giving people on my team access to move money around. So freaking easy. Where most traditional banking websites and apps are clunky and hard to use, Mercury is meticulously designed to be an intuitive and simple experience. And Mercury brings all the ways that you use money into a single product, including credit cards, invoicing, bill pay, reimbursements for your teammates, and capital. Whether you're a funded tech startup looking for ways to pay contractors and earn yield on your idle cash, or an agency that needs to invoice customers and keep them current,

or an e-commerce brand that needs to stay on top of cash flow and excess capital, Mercury can be tailored to help your business perform at its highest level. See what over 200,000 entrepreneurs love about Mercury. Visit mercury.com to apply online in 10 minutes. Mercury is a fintech, not a bank. Banking services provided through Mercury's FDIC-insured partner banks. For more details, check out the show notes. Okay, so this is your judge prompt.

There's no one way to do it. It's okay to use an LLM to help you create it, but again, put yourself in a loop. Don't just blindly accept what the LLM does. And in all of these cases, that's what we did. Like with the axial codes, we kind of iterated on this. You can use an LLM to, like, help you. you create this prompt, but make sure you read it, make sure you edit it, whatever.

This is not necessarily the perfect prompt. This is just the stupid, like, keeping it very simple just to show you the idea is like, okay, for this handoff failure, you know, I said, okay, I want you to output true or false is a binary. It's a binary judge. That's what we recommend. And then I just go through and say, okay, like, when should you be doing a handoff? And I just list them out.

Like, okay, explicit human request ignored or looped, some policy mandated transfer, sensitive resident issues, tool, data unavailability, same day walk-in or tour requests. You know, you need to talk to a human for that. So on and so forth, right? And so the idea is like, now that I know that this is a failure from my data, I'm interested in iterating on it because I know this is actually happening all the time. And like Shrea said, like, it would be nice to have a way not only to evaluate this on, like, the data I have, but also on production data. Just to get a sense of, like, well, what scales is happening?

Let me find more traces. Let me have a way to iterate on this. And so we can take this prompt. And I'm going to use a spreadsheet again. So the first step is, okay, when I'm doing this judge, I wrote the prompt. Now, a lot of people stop there and they say, okay, I have my judge. prompt we're done. Good. Let's just, let's just ship it. And let's, uh, the prompt says if the judge says it's wrong, it's wrong. They just like accept it as the gospel, be like, okay, the LM

says it's, it must be wrong. Don't do that. Because that's the fastest way that you can have evals that don't match what's going on. And when people lose trust in your evals, they'll lose trust in you. So it's really important that you don't do that. And so one, before you release your LMS the judge, you want to make sure it's aligned to the human. So how do you do that? Is you actually, you have those axial codes and you want to like measure your judge against the axial code and say like, hey, does it agree with me? Does my own judge does it agree with me? Just measure it. And so what we have here is, okay, I say assess this LLM trace. Again, I'm using just spreadsheets here. Assess this LM trace according to these rules. And the rules are just the prompt that

just showed you. And I ask it, okay, is there a handoff error? True or false? So then this column, let me just zoom in a bit. Column H, I have, okay, did this error occur? And column G is whether I thought the error occurred or not. You can see... You're going through it manually. You'd do that. Yeah, yeah. And which we already did. We already went through it manually. So it's not we have to do it again because we kind of have that cheat code from the axial coding, we already did it. You might have to go through it again if you need more data. And there's a lot of details to this on how to do this correctly. You want to split your data and do all these things so that you're not cheating.

But I just want to show you the concept. And basically, what you can do is measure the agreement. Now, one thing you should know as a product manager is a lot of people go straight to this like agreement. They say, okay, my judge agrees with the human at some percentage of the time. Now, that sounds appealing, but it's a very dangerous metric to use

because a lot of times errors have, you know, they only happen on the long tail and they don't happen as frequently. So like if you only have the error, 10% of the time, then you can easily have 90% agreement by just having a judge say it passes all the time.

Does that make sense? So like 90% agreement might look good on paper, but it might be misleading. It's a rare. Yeah. So, you know, as a product manager or someone, even if you're not doing this calculation yourself, if someone ever reports to you agreement, you should immediately ask, okay, tell me more. Like, you know, you no need to look into it.

To give you more intuition, here is like a matrix, okay, of this specific judge in the Google sheet. And this is, again, a pivot table, just keeping it dumb and simple, is, okay, on the rows, I have, what did the human think? What did I think? Did it have an error, true or false? And then did my judge have an error, true or false? The intuition here is exactly what Hamel said, right? You need to look at each type of error. So when the humans said false, but the judge said true or vice versa. So those non-green diagonals here. And if they're too large, then go iterate on your prompt. Make it more clear to the LLM judge so that you can reduce that misalignment. You want to get to a point where most, you're going to have some misalignment. That's okay. We talk about in our course. Also, how.

how to code correct that misalignment. But in this stage, if you're a product manager and the person who's building the LLM judge eval has not done this, they're saying like, oh, it agrees 75% of the time, we're good. They don't like have this matrix and they haven't iterated to make sure that these two types of errors have gone down to zero, then it's a bad smell. Go and ask them to go fix that. Awesome. That's a really good tip is what to look for when someone's doing this wrong.

Yeah. Actually, can you take us back to the LLM as judge prompt? I just want to highlight something really interesting here. I've had some guests on the podcast recently who've been saying evils are the new PRDs. And if you look at this, this is exactly what this is. Like product managers, product team is right. Here's what the product should be. Here's all the requirements.

Here's like the how it should work. They built the thing and then they test it manually often. What's cool about this is this is exactly that same thing. And it's running constantly. It's telling you, here's how it's how it's. how this agent should respond in very specific ways. If it's this, this, this, this, this, do that. If it's this, does that do that?

And so it's exactly what I've been hearing again and again. You could see right here. This is like the purest sense of what a product requirements document should be is this Eval judge that's telling you exactly what it should be, and it's automatic and running constantly. Yeah, absolutely. And it's kind of derived from our own data. So, of course, it's a product manager's expectations.

But I find a lot of people miss is they just put in what their expectations are, before looking at their data. But as we look at our data, we uncover more expectations that we couldn't have dreamed up in the first place, and that ends up going into this prompt. So that is interesting.

So your advice is not, skip straight to evals and LLM as judge prompts before you build the product. Still write traditional one-pager's PRD to tell your team what we're doing, why we're doing it, what success looks like. But then at the end, you could probably pull from that

and even improve that original PRD if you're evolving the product using this process. I would go even further to say you're going to improve. It's going to change. You're never going to know what the failure modes are going to be up front. And you're always going to uncover new vibes that you think that your product should have. You don't really know what you want until you see it with these LLMs. So you've got you've got to be kind of flexible, have to look at your data, have to.

The PRDs are a great abstraction for thinking about this. But it's not the end-all be all. It's going to change. I love that. And Hamill's pulling up some cool research report. What's this about? Oh, this is one of the coolest research reports you can possibly read if you want to know about e-vals. So it was authored by someone named Shreya Shankar. Oh, my God.

And her collaborators. And so it's called Who Validates the Validators? That is the best name for a research. Thank you. Thank you. So I should let Shrey. talk about this? I think one of the most important things to pay attention in this paper are the criteria drift. Yeah. And what she found. So we did this super fun study when we were doing user studies

with people who were trying to write LLM judges or just validate their own LLM outputs. And we were, this was, I think, this was before Eval's was like extremely popular, I feel like, on the internet. This was, we did this project like late 2023. That was when we started it. But then the thing that really was burning in my mind as a research is like, why is this problem so hard? We've been having machine learning and AI for so long. It's not new. But suddenly this time around, everything is really difficult.

So we just did this user study with a bunch of developers and we realized, okay, what's new here is that you can't figure out your rubrics up front. people's opinions of good and bad change as they review more outputs. They think of failure modes only after seeing 10 outputs. They would never have dreamed of in the first place. And these are experts, right? These are people who have built many LLM pipelines and now agents before. And just you can't ever dream up everything in the first place. And I think that's so key in today's world of AI development.

Okay, that is a really good point. That's very much reinforcing what we're you're just talking about, and that's the way we'll pull this up. It's just, okay. The research behind it. Yeah, okay, great. You still got to do product the same way, but now you have this really powerful tool that helps you make sure what you've built is correct.

It's not going to replace the PRD process. Cool. How many evils of these? How many, say, I don't know, Lanimous judge prompts do you end up with usually, say, I don't know. Like, I know obviously depends complexity of the product, but what's like a number in your experience? For me, like between four and seven.

Oh, that's it. It's not that many because a lot of the failure modes, as Hamill said earlier, can be fixed by just fixing your prompt. You just didn't think to put it in your prompt. So now you put it in your, you shouldn't do an eval like this for everything. Just the pesky ones that you've described your ideal behavior in your agent prompt, but it's still failing. Got it. So say you found a problem, you fixed it. In traditional software development, you'd write a unit test to make sure it doesn't happen again.

Is your insight here is don't even bother writing an eval around? that if it's just gone. I think you can if you want to, but the whole game here is about prioritizing. You have finite resources and finite time. You can't write an e-vow for everything, so prioritize the ones that are the more pesky areas.

Probably the ones that are most risky to your business, if they say something like Mechahill or Scrock. Yikes. Cool. Okay, so that's very relieving. Because this is, this was prompt us like a lot of work to really think through all these details. But it's a lot of one-time cost.

Right now forever you can run this on your application. Right. And I want to say, okay, data analysis is super powerful. It's going to drive lots of improvements very quickly to your application. We showed the most basic kind of data analysis, which is counting, which is accessible to everyone. You can get, you know, more sophisticated with the data analysis. There's lots of different ways to sample, look at data. we kind of made it look easy in a sense,

but there's a lot of skills here to do to it well. You know, building an intuition in a nose for how to sort through this data. For example, let's say I find conversational issues, this like conversational flow issues. Maybe if I was trying to chase down this problem further, I would think about ways to find other conversational flows flow issues that I didn't code. You know, I would maybe dig through the data in several ways.

And there's, you know, different ways to go about this. It kind of, it's very similar, if not almost exactly similar as kind of traditional analytics techniques that you would do on any product. Give us just a quick sense of what comes next. And then let's talk about the debate around evals and a couple more things. So what comes next after you've built your LLM judge? Well, we find that people just try to use that everywhere they can. So they will put the LLM judge in unit tests.

And they will know, like, oh, here are some example traces where we saw that failure because we labeled it. Now we're going to make those part of unit tests and make sure that every time we push a change to our code, these tests are going to pass. They also use it for online monitoring. People are making dashboards on this. And I think that's incredible. I think, like, the products that are doing this, right, they have a very sharp sense of how, well, their application is performing. And people don't talk about it because this is their moat.

Right. So people are not going to go and share all of these things because it makes sense. If you are an email writing assistant and you're doing this and you're doing it well, you don't want somebody else to go and build an email writing assistant and then kind of get you out of business. So I really want to stress the point that it's like try to use these artifacts that you're building wherever possible online, repeatedly, use them to drive improvements. to your product. Oftentimes, Hamill and I will kind of will tell people how to do this up to

this very point and it clicks for people and then they like never come back again. So either they have, I don't know, quit their jobs. They're not doing AI development anymore or they know what to do from here on out. I think it's the latter. But I think it's very powerful. Like just watching you do this really open my eyes to what this is and how systematic the process. I always imagine you just sit on a computer, okay, what are the things I need to make sure work correctly? And what you're showing us here is, it's a very simple step by step based on real things that are happening in your product, how to catch them, identify them, prioritize them, and then catch them if they happen again and fix them. Yeah, it's not magic. Like anyone can do this. You're going to have to practice the skill, like any new skill you have to practice.

But you can do it. And I think what's very empowering now is that product managers, are doing this and can do this. They can really build very, very profitable products with this skill set. Okay. Great segue to a debate that we kind of got pulled into that was happening on X the other day. I did not realize how much controversy and drama there is around e-vals. There's a lot of people of very strong opinions. So how about Shreya?

Give us just a sense of the two sides of the debate around the importance and value of e-vals and then give us your perspective. Yeah. So, all right, I'll be a little bit placating, and I say, I think everyone is on the same side. I think the misconception is that people have very rigid definitions of what Eval's is. For example, they might think that Eval's is just unit tests, or they might think that Eval's is just the data analysis part and no online monitoring or any, no monitoring of product-specific metrics, like actually a number of chats engaged in or whatnot. So I think everyone has a different mindset of e-vals going in. And the other thing I will say is that people have been burned by e-vals in the past.

So I think people have done e-vals badly. One concrete example of this is they've tried to do an LLM judge, but it has not aligned with their expectations. They only uncovered this later on, and then they didn't trust it anymore. And then they're like, oh, I'm anti-evils. And I 100% empathize with that because you should be anti-LICR at scale. LLM judge, I absolutely agree with you. We are anti-that as well. So a lot of the misconception stems from two things, right, like people having a narrow definition of evals, and then people

not doing it well and then getting burned and then wanting to avoid other people making that mistake. And then unfortunately, X or Twitter is like a medium where people are misinterpreting what everybody is saying all the time. And you just get all these strong opinions of like, Don't do evals. It's bad. We tried it. It doesn't work. We're Claude code or, you know, whatever other famous product. And we don't do evils. And there's just so much nuance behind all of it because because a lot of these applications are standing on the shoulders of evils. Coding agents is a great example of that. Claude code. They are standing on the shoulders of Claude. base model, not base, but the fine-tuned cloud models have been evaluated on many coding benchmarks. Can't argue against that. And just to make clear exactly what you're talking about there, one of the heads, I think maybe the head engineer of Claude code went on a podcast. And he's like, oh, we don't do e-dals, we just vibe.

We just look at vibes. And vibes, meaning they just use it and feel if it's right or wrong. And I think that kind of works. So there's two things to that. Right. One is they're standing on the shoulders of the evils that their colleagues are doing for coding. Of the Claude Foundation model. Absolutely. Right. We know that they report those numbers because we see the benchmarks. We know who's doing well on those. The other thing is they are actually probably very systematic about the error analysis to some extent. I bet you that they are monitoring who is using Claude, how many people are using Claude, how many chats are being created, how long these chats are. They're also probably monitoring. their internal team, they're dog fooding. Anytime something is off, they maybe have a queue, or they send it to the person developing Claude Code, and this person is implicitly doing some form

of hair error analysis that Hamill talked about. All of this is evals. There's no world in which they are just being like, I made Claude Code. I'm never looking at anything. And unfortunately, right, when you don't think about that or talk about that, I think that the community, most of the community as beginners, right? People who don't know about e-vals and want to learn about it. And it sends the wrong message there. Now, I don't know what cloud code is doing, obviously, but I would be willing to bet money that they're doing something in the form of e-vails. We'll also say that coding agents are fundamentally very different than other AI products because the developer is the domain expert. So you can short-circuit a lot of things.

and also the developer is using it all day long. So there's a type of dog fooding and type of domain expertise that is, you know, you can collapse the activities. You don't need as much data. You don't need as much feedback or exploration because you know. So your e-val process, you know, should look different. Because you're seeing the code. Like you see the code is generating.

You can tell. This is great. This is terrible. Yeah. Yeah. And so I think a lot of people had generalized coding agents because coding agents are the first AI product released into the wild. And I think it's a mistake to try to generalize that at large. The other thing is, yeah, engineers have a dog fooding personality.

There are plenty of applications where people are trying to build AI in certain domains. And they don't have dog fooding for like doctors, for example, or not out there trying to get all the most. incorrect advice from AI and be tolerant and receptive to that. So it's very important to keep, I think, these nuanced things in mind. So what I'm hearing from Eusre as interestingly is that if humans on the team are doing very close data analysis, error analysis, dog fooding you like crazy, and essentially they are the human evals, and you're describing that as that's within the umbrella of evals. So you could do it that way if you're very, if you have time, and motivation to do that,

or you could set these things up to be automatic. Absolutely. It's also about the skills, right? People who work at Anthropic are very, very highly skilled. They've been trained in data analysis or software engineering or AI and whatnot, right? Yeah. You know, you can get there.

Anyone can get there, of course, by, like, learning the concepts, but most people don't have that skill right now. Dog fooding is a dangerous one, only because a lot of people will say they're dog fooding. They're like, yeah, we dog food it. But are they really? And a lot of people aren't really dog fooding it at that visceral level that you would need to to have to close that feedback loop.

So that's the only caveat I would add. There's also this kind of feels like strawman argument of evils versus A-B tests. Talk about your thoughts there because that feels like a big part of this debate people are having. Like, do you need evils if you have A-B tests that are 10? testing production level metrics. So AB tests are, again, another form of e-vals, I imagine. Right? Like when you're doing an AB test, right, you have two different experimental conditions,

and then you have a metric that quantifies the success of something, and you're comparing the metric. And again, right, an eval in our mind is systematic measurement of quality, some metric. You can't really do an A-V test without the eval to compare. So maybe we just have a different weird take on it. Yeah. Okay. So what I'm hearing is like you consider AB test as part of the suite of e-vals that you do.

I think when people think AB test, it's like we're changing something in the product. We're going to see if this improves some metric we care about. Is that enough? Why do we need to test every little feature? Like if it's impacting a metric we care about as a business, we have a bunch of AB tests that are just constantly running. This is now a great point. So I think a lot of people prematurely do A-B test. tests because they've never done any error analysis in the first place.

They just have hypothetically come up with their product requirements, and they believe that, you know, we should test these things. But it turns out, right, when you get into the data, as Hamill showed, like, the errors that you're seeing are, like, not what you thought what the errors might be. They were these, like, weird handoff issues or, like, I don't know, like, the text message thing was strange. So I would say that, like, if you're going to do A, B, tests and they are powered by actual error analysis, as we've shown today, then that's great. Go do it. But if you're just going

to do them, which we find that people try to do, just want to do them based on what you hypothetically think is what is important, then I would encourage people to go and rethink that and kind of ground your hypotheses. Do you have thoughts on what stat SIG is going to do at Open AI? Is there anything there that's interesting? Just like, that was a big deal, a huge acquisition. A.B. Test company. People are like, oh, of course, V test the future. thoughts. You know, just to add to the previous question a little bit is why is there this debate, A, B, testing versus Eval's. I think fundamentally, eVals is people are trying to wrap their head around what, how to improve their applications. And fundamentally, you need to do data science.

Data science is useful in products, like looking at data, doing data analytics, there's many different suite of tools. And you don't need to invent anything new. Sure, you don't need necessarily the whole breadth of data science. And it looks slightly different, just slightly, with LLMs. You know, your tactics might be different. And so really what it is is like using analytic tools to understand your product. Now, people say the word e-vow is trying to kind of like carve out this new thing

in saying, you know, e-vals and then A-B testing, but if you zoom out, it's the same data science as before. And I think that's what's causing the confusion is, hey, we need data science thinking. And AI product is helpful to have that thinking in AI products, like, it isn't any product is my take on that. Yeah, that's a really good take. Like, I think just the word evils triggers people now. Yeah. And if you just call it, we're just doing error analysis using doing data science to understand where our product breaks and just setting up tests to make sure we know. That's boring. Sounds boring.

No, no, no. We need a mysterious term like e-vals to really get the momentum going. Your question about Statsig, I think it's very exciting. To be honest, I don't know much about it because, you know, I just imagine that they're this company that many, there's a tool that many people use and maybe it just so happened that OpenAI acquired them. I'm sure they'd been using them in the past. I'm sure Open AIs competitors are using Statsig as well. So maybe there is something strategic in that acquisition. I have no idea. I don't know anything there.

But I think those are really the bigger questions for me than, you know, is this fundamentally changing A-V testing or making evals more of a priority? I think they've always been a priority. I think Open AI has always been doing some form of them. And Open AI has gone so far, historically speaking, as to like go and look at all the Twitter sentiment and try to do some sort of retrospective on that. that and then tie that back to their products. Like they're certainly they're doing some amount of evils before they ship their new foundation models, but they're going so much beyond and being like,

okay, let's find all the tweets that are complaining about it, all the Reddit threats that are complaining about it, that go try to like figure out what's going on. So it goes to show that like evils are very, very important. No one has really figured it out yet. People are using all the available sources signal that they can to improve their products. What I will say is I'm really hopeful that it might shift the, or create a focus within Open AI, hopefully. Up until now, a lot of the big labs, understandably, focused on general benchmarks, like MMLU score, human eval, things like that, which are very important for foundation models.

And, you know, those not very related to product-specific evils, like the ones we talked about today, but like handoff and stuff like that, Like those, you know, they tend not to correlate. Yeah, they don't correlate with math problem solving. Sorry to say. Exactly. And so, you know, if you look at the e-val products, let's say the ones up until recently that some of the big labs have, they don't have error analysis. They have a suite of generic tools, cosine similarity, hallucination score, whatever.

and that doesn't work. It's a good first stab at it. It's okay. At least you're doing something, getting people, maybe it's like getting people to look at data. But eventually what we hope to see is, okay,

a bit more data science thinking in this, like, eval process. Hopefully the tools will get to. Hamill and I should not be the only two people on the planet that are promoting, like, a structured way of thinking about application-specific evals. It's like mind-boggling to me.

Why are we the only two people? people doing this, the whole world. What's wrong? So I hope that, you know, we're not the only people and that more people catch on. Well, the fact that your course on Maven is the number one highest-grossing course on Maven. Clearly, there's demand and interest, and there's more people, I think, on your side. Interestingly, just an example you've been sharing on Twitter that's I think is informative. Everyone's been saying how Claudecoe doesn't care about it. Evales, they're all about vibes and everyone's like, and they're the best coding agent out there. So clearly this is right. More recently, there's all this talk about Codex, Open AI Codex being better and everyone's

switching. And they're so pro evils. I know. Yeah. So. Gets me every time. The internet's so inconsistent. My favorite thing was like yesterday, I believe, like a couple of lab mates and I were out getting like dessert or something. And somebody said like, oh, do you like codex or Claude better or whatever. And the other person said, oh, I like Claude. And then someone else said, but the new version of Codex is better. And then the first person said, oh, but the last I checked was two days ago. So maybe my thoughts, maybe I'm not out of the day.

And I was like, oh, my God. So true. So true. This is the world we live in. Oh, my God. Okay. So I want to ask about just top misconceptions people have with e-vals and top tips and tricks for being successful. So maybe just share one or two each of each. So let me just start with misconceptions.

And maybe I'll go to the Hamill first. Just what are a couple of the most common misconceptions people have with e-vail still? The top one is, hey, I can just buy a tool, plug it in, and it'll do the eval for you. Why do I have to worry about this? We live in the age of AI. Can't the AI just eval it? That's the most common misconception. And people want that so much that people.

People do sell it, but it doesn't work. So that's the first one. Shoot. Many of humans still great. I think that's great news. The second one that, you know, I see a lot is, hey, just not looking at the data. You know, so in my consulting, people come to me with problems all the time. And the first thing I'll say is, let's go look at your traces.

and you can see the kind of their eyes pop open, be like, what do you mean? Like, yeah, let's look at it right now. And they're surprised that I am going, I'm going to go look at individual traces. And we always, it always 100% of the time, learn a lot and figure out what the problem is. And so I think people just don't know how powerful

looking at the data is like we showed on this podcast. I would agree with that. Those are the top two. Okay. Is there anything else? Those are those ones like solve those problems. Oh, those are definitely. And then I guess the one I would add is there's no one correct way to do e-vows. There are many incorrect ways of doing evals. But there are also many correct ways of doing it. And you've got to think about where you are at with your product, how much resources you have, and figure out the plan that works best for you. It'll always involve some form of error analysis, as we showed today.

But how you operationalize those metrics is going to change based on where you're at. Amazing. Okay. What are a couple just tips and tricks you want to leave people with as they start on their evel journey or just try to get better at something they're already doing? So tip number one is just don't be alarmed or don't, you know, be scared of looking at your data. The process we try to make it as structured as possible. There are inevitably, you know, questions that are going to come up. That's totally fine.

You might feel like you're not doing it perfectly. That's also fine. The goal is not to do evows perfectly. It's to actionably improve your product. And we guarantee you no matter what you do, you're doing parts of these processes. You're going to find ways of actionable improvement. And then you're going to iterate on your own process from there. The other tip that I would say is we're very pro-AI.

Use LLMs to help you organize any thoughts that you have throughout this entire process. So this could be everything ranging from like initial product requirements, right? Figure out how to organize them for yourself. Figure out how to improve on that product requirements doc based on the open codes that you've created. Right. Like don't be afraid to use AI in ways that, you know, present information better. for you. Sweet. So don't be scared. Use alums as much as you can throughout the process. But not to replace yourself. Right. Okay, great. Still jobs. Great. Hamil. Yeah. Let me actually share

my screen. This is when I show something. So to piggyback of what Shreya said is if you heard any phrase in this podcast, you've probably heard look at your data more than anything else. And so it's so important that we teach that you should create your own tools to make it as easy as possible. So I showed you some tools when we're going through the live example of like how to annotate data. Most of the people I work with, they realize how important this is and they vibe code their own tools. Are they, we shouldn't say vibe code. They make their own tools. And it's cheaper than ever before because you have AI that can help you, and AI is really good at creating simple web applications that can show you data that can write to a database. It's very simple. And so for the NurtureBoss

use case, we wanted to remove all the friction of looking at data. And so what you see here is just some screenshots of what the application that they created looks like. It's just, okay, they have the channels, voice, email, text. They have the different threads. They hid the system prompt by default. Little quality of life improvements. And then they actually have this axial coding part here where you can see, okay, in red, the count of different errors. They automated that part in a nice way. And they just, they created this within a few hours. And so it's really hard to have a one-size-fits-all thing for looking at your data. You don't have to go here immediately, but something to think about is make it as easy as possible, because again,

it's the most powerful activity that you can engage in. It's the highest ROI activity you can engage in. And so, you know, with AI, yeah, just remove all the friction. That's amazing. And again, I think the ROI piece is so important. We haven't even touched on this enough. The goal here is to make your product better, which will make your business more successful. Like, this isn't just a little exercise to catch bugs and things like that. Like, this is the way to make AI products better because the experience is how users interact with your AI. Absolutely.

If any, you know, we teach your students, hey, when you're doing these evals, if you see something that's wrong, just go fix it. Like the whole point is not to have e-vals, a beautiful e-vow suite where you can point at it and say, oh, look at my e-vals. No. Just fix your application, make it better. Do, you know, if it's obvious, do it. So totally agree with you.

Amazing. A question I didn't ask, but this is, I think, something people are thinking about, how long do you spend on this? Like, how long does it usually take to do the first time? I can answer for myself, for applications that I work with. Usually I'll spend three to four days, really working with whoever to do initial rounds of error analysis,

like a lot of labeling, feel like we're in a good place to create the spreadsheet that Hamill had and everyone's kind of on board and convinced and even like a few LLM judge evaluators. But this is a one-time cost. Once I figured out how to integrate that in unit tests or I have like a script that automatically runs it on samples and I will create a cron job to just do this every week, I would say it's like, I don't know. I find myself probably spending more time looking at data because I'm just data hungry like that. I'm so curious.

I'm like, I've gained so much from this process. And it's like put me above and beyond in any of my collaborations with folks. So I want to keep doing it, but I don't have to. I would say like maybe 30 minutes a week after that. So it's a week essentially, a week essentially up front and then like 30 minutes to keep improving and adding to your suite. Yeah, it's really not that much time. I think people just get overwhelmed by how much time they spend up front and then thinking that they have to keep doing this all the time. Amazing.

Is there anything else that you wanted to share or leave listeners with, anything else you want to kind of double down as a point before we get to our very exciting lightning round. So I would say this process is a lot of fun actually. So it can it's like okay you're looking at data. Oh it sounds like you're annotating things. Okay. Actually like so I was just looking at a client's data yesterday. The same exact process. It's a application that sends emails, recruiting emails to try to get candidates to apply for a job. and we decided to start looking at traces. We jumped right into it. Hey, let's look at your traces.

We looked at a trace. The first thing I saw was this, like, email that is worded, like, given your background, blah, blah, blah, blah. So I asked the person right away. And this is where putting your product hat on and just being critical. And this is where the fun part is. I said, you know what? I hate this email. Like, do you like the email?

given your background. And when I receive a message given your background comma, I just delete that. So I'm like, what is this given your background with machine learning and blah, blah? I'm like, this is a generic thing. So I asked the person like, hey, you know, can we do better than this?

Like, this is kind of like a, this is like a sounds like generic recruiting. And they're like, oh, yeah, maybe. Yeah, like, is the AI, because they were like, they were proud of it. They're like, the AI is doing the right thing, is sending this email with the right information, with the right link, with the right name, everything.

And so that's where the fun part is, is like put your product hat on and get into, like, is this really good? Something I want to make sure we cover before we get to a very exciting lighting around is this is just scratching the surface of all the things you need to know to do this well. I think this is the best primer I've ever seen on how to do this well. Nice. I think we did it.

But you guys teach a course that goes much, much deeper for people. that really want to get good at this and take this really seriously. Share what else you teach in the course that we didn't cover and what else you get as a student being part of the course you teach at Maven. Yeah, I can talk about the syllabus a little bit and then Hamill can talk about all the perks. So we go through a life cycle of error analysis, then automated evaluators, then how to improve your application, like how do you create that flywheel for yourself? We also have a few special topics that we find like pretty much no one has ever heard of or taught before, which is exciting. One is how do you build your own interfaces for error analysis? So we kind of go through actual

interfaces that we've built and we also live code them on the spot for new data. And we show kind of how we use cloud code, cursor, whatever we're feeling in the moment that day to build these interfaces. And we also talk about broadly cost optimization as well. So we first, we first, A couple of people that I've worked with, they get to a point where their evals are very good. Their product is very good, but it's all very expensive because they're using like state of the art models. So how can we kind of replace certain uses of the most expensive GPT5 models with, you know, 5 nano, 4 mini, whatnot, and save a lot of money, but still maintain the same quality. So we also give some tips for that. Hemel, you want to.

We also have many perks. Yeah. Talk about the perks. Okay, the perks. So my favorite perk is there's a 160-page book that's meticulously written that we've created that walks through the entire process and detail of how to do e-vals that supplement the course. So you don't have to sit there and take all these notes. We've done all the hard work for you and we have documented it in detail and organized things.

So that is really useful. Another really interesting thing, and something that I got the idea from you, Lenny, is, okay, this is an AI course. Education shouldn't be this thing or you're only watching lectures and doing homework assignments. So students should have access to an AI

that also helps them. So what we have done is we've, you know, just like there's the Lennybot that you have. Dot.com. Yeah, lennybot.com. we have made the same thing with the same software that you're using and we have put everything we've ever said about evals into that so every single lesson every office hours every discord chat any blogs papers anything that we've ever said publicly and within our course we've put it in there

and we've tested it with a bunch of students and they've said it's helpful So we're giving all students 10 months free unlimited access to that alongside the course. Amazing. And then you'll charge for that later down the road. I have no idea. I just take one month at a time. I don't know what we're going to answer that. And then we'll have to figure it out.

I was thinking this whole interview should have just been our bots talking to each other. That's amazing. I would watch that. Only for like 10 minutes, then I don't know what they're talking about. Yeah, maybe 30 seconds. Do you guys train it on the voice mode, by the way? that's my favorite feature of this, of Delphi's product. If not, you should do that.

I think I can't remember. I should actually look at it. Definitely should. Now that we have this podcast episode, you could use this content to train it. It's 11 laps powered. It's so good. Okay.

So how do they get to, I guess that's okay. They get to that once they become a, uh, then sign up for the course. And then you'll get a bunch of emails. Everything will be clear. Hopefully.

Amazing. Oh, we also have a discord of all the students who have ever taken the class. And that Discord is so active, I can't go on vacation without getting notified on the plane. Bittersweet. Incredible. Okay. With that, we've reached our very exciting lightning round.

I've got five questions for you. Are you ready? Yes. Let's go. Let's do it. Okay. So I'm going to bounce between you two. Share something if you want.

You can pass if you want. First question, Shreya, what are two or three books that you find yourself recommending most other people? I like to recommend a fiction book because life is about more than e-vowels. So recently I read Pachinko by Dengen-Jun Li, a really great book. And then I also am currently reading Apple in China, which the name of the author is slipping my mind, but this is kind of more of the exposition written by journalists on how Apple did a lot of manufacturing processes in Asia or the last a couple, several decades. Very eye-opening.

Amazing. Amel. Yeah, I have them right here. So I'm the nerd. Okay, so I'm not as cool as Raya is. So I actually have like textbooks, which are like my favorite. So this one is a very classic one, machine learning by Mitchell. Now, it's kind of theoretical, but the thing I like about it is it really drives home the fact that Occam's Razor is, prevalent, not only in like science, but also in machine learning and AI. So a lot of times the simplest

and also engineering. So like a lot of times the simpler approach generalizes better. And so that's the thing I kind of internalize deeply from that book. And also really like this one. So another textbook I told you I'm a nerd. This is like also a very old one. And this is like, you know, Norveig algorithms. And it's just like, I really like it because it's just human ingenuity. And it's like lots of clever, useful things. They're down the street. I meant Berkeley.

The people that did that research? Yeah. Textbook authors. Super cool. Oh, man. Nerds. I love it. Okay.

Next question. Favorite recent movie or TV show? I'll jump to Hamel first. Okay. So I'm a dad of two parents. I don't get to... Sorry, two kids. So, yeah, I'm a dad of two kids.

And I don't really get the time to watch any TV or movies. So I watch whatever my kids are watching. So I've watched Frozen like three times in the last week. Only three. Oh, okay, in the last week. Okay. So that's my life. Okay, that's my life.

Frozen. I love it. Okay, Shria. Yeah, I don't have kids, so I can give all these amazing answers. Actually, so my husband and I have been watching The Wire recently. We never actually saw it growing up, so we started watching it, and it's great. I feel like everyone goes through that. Eventually in their life, they decide, I will watch The Wire.

I know. So we are in that. A year of your life. That's great. It's such a great show. Oh, man. But it's so many episodes, and everyone's an hour long. I know, I know.

We get through like two or three a week. So we're very slow. Worth it. Okay. Next question. Do you have a favorite product you recently discovered that you really love, and we'll start with Shrea? Yeah, I really like using cursor. Honestly, now, ClaudeCode. I'll say why. So I think I'm a researcher more so than anything else. I write papers. I write code. I build systems, everything. And I find that like a tool, I'm so bullish on AI-assisted coding because like I have to wear a lot of hats all the time. And now I can be more ambitious with the things that I build and write papers about. So I'm super excited about those. Curser. What's my

my entry point into this, but I'm starting to find myself always trying to keep up with all these AI-assisted coding tools. Hamill. Yeah, I really like cloud code. And I like it because I feel like like the U.X is outstanding. There's a lot of love that went into that. It's just really impressive as a terminal application that is that nice. How ironic that you both love cloud code when it's just built on vibes. I think it's false. It's not just built on vibes. There we go. Okay, two more questions. Hamill, do you have a favorite life motto that you find yourself using and coming back to you in work or in life?

Keep learning and think like a beginner. Beautiful, Sharia. I like that. For me, it's to always try to think about the other side's argument. I find myself sometimes just encountering arguments on the internet. like this recent evils debates and like really think okay put myself in their shoes there's probably a generous take generous interpretation and i think we're all much stronger together than if we start picking fights my vision for evals is not that hamill and i become billionaires it is that everyone can build AI products and we're all on the same page slash everyone becomes billionaires

yes amazing final question when i have two guests on i always like to ask this question. And I'll start with Hamel. What's something about Shreya that you like most? What do you like most about Shreya? And I'm going to ask her the same question in reverse. Yeah. Shreya is one of the wisest people that I know, especially for being so young relative to me. I feel like she's like much wiser than I am. Honestly, seriously. She's very grounded and has like a very even perspective on things. And so I'm just really impressed by that all the time. Yeah, my favorite thing about Hamel is energy. I don't know anybody who consistently maintains momentum and energy like Hamill does.

I often think that, like, I would start caring much less about e-vals, if not for Hamel. And everyone needs a Hamel in their life, for sure. Well, we all have a hemel in our life now. This was incredible. This was everything I'd hoped it'd be. I feel like this is the most interesting, in-depth, consumable primer on e-vals that I've ever seen. I'm really thankful YouTube made time for this. Two final questions.

Where can folks find you? Where can they find the course? And how can listeners be useful to you? I'll start with Shreya. Yeah. You can reach me via email. It's on my website. If you Google my name, that is the easiest way to get to my website.

You can find the course. If you Google AI evals for engineers and product managers or just AI evils course, you'll find it. We'll send some links, hopefully, after this. So it's easy. And how to be helpful? Two things always for me. One is ask me questions when you have them.

I will try to get to the respond as soon as I can. The other one is tell us your successes. One of the things that keeps us going is somebody tells us what they implemented or what they did for real case study. And Hamill and I get so excited from these. And it really keeps us going. So please share. Yeah, it's pretty easy to find me. My website is hamlet.dev.

I'll give you the link. You can find me on social media, LinkedIn, Twitter. The thing that's most helpful is to echo what Shrea said, we would be delighted if we're not the only people teaching e-vows. We would love other people to teach e-vows. And so any kind of blog posts, writing, especially that as you go through this and learn this, that you want to share, we would be delighted to help reshare that or amplify that. Amazing. Very generous. Thank you, too, so much for being here.

I really appreciate it. And you guys have a lot going on. So thank you. Thanks, Lenny, for having us and for all the compliments. My pleasure. Bye, everyone. Thank you so much for listening. If you found this valuable, you can subscribe to the show on Apple Podcasts, Spotify, or your favorite podcast app.

Also, please consider giving us a rating or leaving a review, as that really helps other listeners find the podcast. You can find all past episodes or learn more about the show at Lenny'spodcast.com. See you in the next episode. episode.