The AI Daily Brief: Artificial Intelligence News and Analysis - Your Company Doesn’t Need an AI Strategy

Episode Date: June 19, 2026

...it needs an AI learning system. This episode argues that the Fable 5 disruption exposed a deeper enterprise problem: companies can’t treat AI as a vendor strategy. The real advantage will come from building learning systems that capture institutional judgment, workflow traces, private evals, and model-portable IP. In the headlines: could Anthropic and the White House be headed for a resolution? Register for our new enterprise-grade AI training programs: http://training.besuper.ai/⁠Brought to you by:KPMG – Research from KPMG and the University of Texas at Austin shows the highest-impact AI users treat AI like a reasoning partner — and those skills can be taught at scale. Learn more at ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠kpmg.com/us/Sophisticated⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Section - Section turns AI investment into workforce transformation and ROI - ⁠⁠⁠https://www.sectionai.com/⁠⁠⁠Outsystems - Stop wondering how AI will change your business and start building the agents that will lead it - ⁠⁠http://outsystems.com/⁠⁠Scrunch - The AI customer experience platform - ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://scrunch.com/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Zenflow Work - Agents for knowledge work - ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://zenflow.free/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Blitzy - Want to accelerate enterprise software development velocity by 5x? ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://blitzy.com/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠MissionCloud - Eliminate AWS complexity with end-to-end cloud and AI services ⁠⁠https://www.missioncloud.com/⁠⁠AssemblyAI - The best way to build Voice AI apps - ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.assemblyai.com/brief⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Robots & Pencils - Cloud-native AI solutions that power results ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://robotsandpencils.com/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠The AI Daily Brief helps you understand the most important news and discussions in AI. Subscribe to the podcast version of The AI Daily Brief wherever you listen: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://pod.link/1680633614⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Our Newsletter is BACK: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://aidailybrief.beehiiv.com/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Interested in sponsoring the show? [e-mail]

Today on the AI Daily Brief, you don't need an AI strategy. You need an AI learning system. Before that, on the headlines, might we finally be heading for resolution between the White House and Anthropic? The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. All right, friends, quick announcements before we dive in. First of all, thank you to today's sponsors, KPMG, robots and pencils, mission cloud, and out systems. To get an ad-free version of the show, go to patreon.com slash AI Daily Brief, or you can subscribe in Apple Podcasts. And if you want to learn more about sponsoring the show, you can send us a note at Sponsors

at AIDily Brief.Ai. Now, one other thing that I wanted to mention today, we have had a ton of interest in participation in the agentic work learning programs that we've been doing. But we wanted to make it even more suited for this particular moment and for the needs of the enterprise. With that in mind, Enterprise Claw has a new name, which is the executive agent leadership program. We took everything that has been working from the first three cohorts, the Build First

approach, the Enterprise Playbooks, tool agnostic architecture, and evolved it for where things are right now. Think token economy, vendor resilience, security governance, multi-agent orchestration. Basically, the stuff that enterprises are actually thinking about today. The program is still being led by Newfar Gas Bar, but is now being offered in partnership with super-intelligent to bring this to enterprise teams at scale with a sharper focus on what enterprises need. That means security first, bring your own tools, and playbooks your organizations will actually run on. We actually now have two programs under one umbrella, the executive catch-up four-week sprint. that's where you go from using AI casually to running a personal AI system,

and the executive agent leadership program that is six weeks, formerly enterprise claw, that is where you go deep. You build enterprise agent systems, wire them to real tools and data, and produce the governance and scaling playbooks your org needs. Now, if you are interested in that program, the next cohort launches June 29th, and you can find all of the information about this at training.bysuper. That's training.b super.aI.

Might we actually be heading into the end of the week with some good news, on the horizon? New reports suggest that talks between the White House and Anthropic are actually moving forward in something of a positive direction. Specifically, Politico reports that the talks have shifted towards designing a framework to assess the severity of security flaws and AI models. By way of background, heading into the week, Anthropics' position was that the fable jailbreak wasn't a serious issue and that the administration was overreacting based on a misunderstanding. It was clear from the reporting that the administration did not like this disposition. First of all, from a technical perspective, they wanted Anthropic to, quote-unquote, fix the jailbreak and wasn't willing to budge on that.

But second, they didn't like the attitude which they saw as nonchalance, colored by more than a bit of hypocrisy as well, given Anthropics brand as the safety-focused AI lab. Now, however, writes Politico, the negotiations between Anthropic and the administration reflect an understanding that no AI model can be completely immune to hacking, part of Anthropics' initial defense of its model, and that government should lay out the rules for companies to measure security risks by. Politico adds that although the export controls haven't been lifted at this stage, the shift towards setting technical standards is a sign that talks are progressing. As recently as a couple of days ago, the tone was very different, with the White House seemingly digging its heels in based, it seems, on a lack of technical understanding.

There has also been mounting pressure on the government to find a way to back down. Throughout the week, this type of use of export controls has been criticized by everyone from security experts to business leaders to, to of course, the AI community. In another story, Politico argued that the fable ban might be able to, not even be legal. They note that the Commerce Department has yet to file the proper paperwork to support the use of export controls, with former Commerce Department official Kevin Wolf suggesting the heavy-handed approach wouldn't scale. If this is going to be their position going forward with every other model or every other data center, that will be a dramatic shift.

And as for the specific case of blocking foreign nationals from logging into a cloud service, Wolf added, I don't know what the legal authority is for that. Speaking to the verge, Andrew Reddy, a professor of public policy at UC Berkeley, said that this was an unsettled area of law to say the least. In some ways, he said, I think this episode makes clear the unsustainability of the existing governance regime. If creating models that are impossible to jailbreak becomes the de facto standard for the United States, then it will have no AI models. Now, overnight, the New York Post added some further details on what Anthropic has offered. Sources said that Anthropic has pledged to work more closely with the White House, improve communication with the administration, and resolve

security concerns more quickly moving forward. The Korean press also published a quote from Anthropic managing director for International, who during a press conference in Seoul on Wednesday said, we are very confident that in the coming days the models will become available again. Now, I will say I have a bit of a concern that we all want Fable Back so badly, that we're overreading the first positive signals we've gotten in a week as the impending resolution of a thing which might still have a ways to go, but of course we can hope. Many are, to put it mildly, unimpressed with the way that things are resolving. Zach Corman wrote, who is leading this from the White House's side?

Or is it just Pete Hegseth and Howard Lutnik sitting with Anthropics engineers writing evals? Kevin Bankston points out, this is what the Center for AI Standards and Innovation is supposed to be for, to develop open standards. But hey, I'm sure Anthropics competitors will love them secretly co-designing de facto government standards with the White House instead. Aaron Levy from Box thinks we're getting a preview of what the governance regime is going to look like from here on out. He said that while these frameworks might seem small, the, quote, implications are massive.

It'll mean he writes that each model, update will go through an extensive review, testing, and feedback process. And in that process, lots of groups will weigh in on the risk of the model, and there will be lots of subjectivity on what the actual risks are or practicalities of exploiting those risks. He argues that if nothing else, this could significantly change the way we get model releases, moving away from this paradigm of quick iterative releases, to something that's much more irregular with bigger updates at one time. Now, later in the evening, another story broke, which may or may not be related to all of this. Bloomberg reports that Commerce Secretary Howard Lutnik last night told ASML,

that the U.S. government believed that one of its ultraviolet lithography or EUV machines may have somehow made its way into China. Senior administration officials argued that they have evidence that ASML is, in their words, not acting in good faith. Now, this is a developing story, but could be quite a big deal when it comes to the U.S.-China AI balance. Now, staying in Washington, Bernie Sanders, has unveiled his plan to create a $7 trillion, yes, that's trillion with a T-dollar sovereign wealth fund by taxing AI companies. In newly unveiled legislation, Sanders laid out the full legal framework for the fund, which would, for some context, be larger than the Social Security Trust Fund. Funding would be provided via a one-time 50% tax on the equity of the largest AI companies.

The legislation defines this as any company with more than 200 million in annual AI sales. That would, of course, include many public companies and a deep list of startups way beyond open AI and anthropic. That is how Sanders estimates the tax would raise $7 trillion. On the governance side, the fund would be managed by an independent commission appointed by the president and confirmed in the Senate, similar to political appointments to government departments. The fund would also hold voting shares, so the commission would have the ability to appoint board members and influence corporate decision making. The bill summary states this commission should use their power to, quote, block decisions that hurt the American people and to push for policies

that help them. Sanders proposes that the fund should distribute a 5% annual dividend in the form of direct payments of more than $1,000 annually to every American. And if the company's held by the fund grow, which I can't imagine, given that they are effectively being run by the government, Sanders is proposing that excess return should be used to fund public goods, including education, housing, and health care. Figuring out some magic that has somehow eluded markets for the last 500 years, Sanders also said the taxpayers would be insulated from losses, commenting, we're not going to lose any money even if there's a burst in the bubble. Summing up the legislation, Sanders said,

what this bill does is not complicated. It gives the American people the ability to prevent AI developments which will negatively impact their lives. And while some have advocated for some form of public ownership of AI companies, Sanders acknowledged his bill goes a little farther than most had in mind. Sanders commented, I think people like Sam Altman and Trump, who may be sympathetic to this, are saying, okay, look, we're making zillions of dollars so we're going to be nice guys and maybe we'll buy off the public.

We'll give 5% of our profits back into the government. That's not what we're talking about. What we're talking about are two very different things. I think when it comes to any proposal like this, you have to view it in two ways. The first is the very literal interpretation. Bernie Sanders isn't dancing around the fact that this is effectively nationalization of AI.

Now, in his mind, that's to make sure that AI benefits the American people, but that's still what it amounts to. 50% of any company with over $200 million in revenue, which also means de facto control because they're voting shares. That is nationalization of the entire AI industry. Now, I hope outside of acknowledging that that is literally what they are going for here, we don't have to spend all that much time actually debating the merits of this bill as it's written. The second and more important way to look at this is how it intersects with the broader discourse. Vice President J.D. Vance actually discussed the Bernie plan in a podcast interview this week. Now, the agreement here is Vance saying the president likes the idea as sort of a sovereign

wealth fund idea of the United States taking some stake in these AI companies. The but is, Vance says, the model where you just take from some people and give to other people has never provided a stable society. You've got to give workers a seat at the table. Now, interestingly, and showing just how topsy-turvy AI makes everything, Vance went on to discuss supporting labor unions in this. The point is, while you might be able to be able to be able to be to rely on folks like Bernie Sanders to create the ten poles of where the AI policy discussion is going to be, I don't think applying a traditional left-right lens to this is really going to be how it plays out. Moving over to the business world, Accenture got hammered this week on weak

earnings as the market priced in AI disruption. The consulting giant reported a 2% drop in bookings for the third quarter and a reduction of revenue forecasts falling below analyst's expectations. That was enough to send the stock tumbling by 18% on Thursday, reaching its lowest level in almost a decade. The stock has now been cut in half this year. Now, during the earnings call, Accenture blamed the Iran War for their miss, stating the company now has a $400 million hole in their Middle East business. However, many are pointing to their lack of performance in guiding the AI transformation. Pat Petiti, the CEO of a rival AI consulting platform, said, real AI implementation requires deep domain expertise in the function where AI will actually be used. That's exactly what they

lack and investors are noticing. CEO Julie Sweet did an interview with CNBC in which she basically said that their investors are missing the fact that Accenture is positioned to get an AI tailwind as companies expand their transformations. Mostly the response of the internet was snarky, with Greg on X summing up the vein of tweets, lots of people say AI isn't actually good enough to replace people yet, but most of them haven't hired Accenture before. Last of today, two new feature announcements to check out if you have access to them. Claude Code has announced artifacts, which in many ways is similar to Codex's sites that we profiled a couple weeks ago. They describe it as interactive pages built from your sessions, like a PR walkthrough or a living project dashboard shared

with your team at a private link. This is part of the trend of moving AI from a single player to a multiplayer experience, and it's currently available on team and enterprise plans. Meanwhile, Codex's Thursday update was a new feature called Record and Replay that they describe as letting you show Codex a recurring task, like filling an expense report or submitting a time-off request, which Codex then turns that demo into an inspectable, editable skill. This one has a ton of people excited. With Jason from OpenAI writing, boy, is it a bad day to be a manual workflow that crosses application boundaries on your computer. Microsoft's Nicholas Bustamante points out that this is especially relevant for organizations dealing with legacy systems, writing, and that's why computer use

and browser use are so important. You can record actions on people's computers and the AI will do the work end to end. This is crucial for old software with no APIs. Fun to see the harness feature set continue to evolve even as the models are a little stuck in purgatory, but for now that that's going to do it for the headlines. Next up, the main episode. One of the most important AI questions right now isn't who's using AI. It's who's using it well. KPMG in the University of Texas at Austin just analyzed 1.4 million real workplace AI interactions and found something surprising. The highest impact users aren't better prompt engineers. They treat AI like a reasoning partner. They frame problems, guide thinking, iterate, and push for better answers. And the good

news, these behaviors are teachable at scale. If you're trying to move from AI access to real capability, KPMG's research on sophisticated AI collaboration is worth your time. Learn more at KPMG.com slash US slash sophisticated. That's KPMG.com slash us slash sophisticated. I cover the capability gap between AI potential and AI reality every day on the show. Most companies are still figuring out how to start. Robots and pencils is already launching and scaling. Agendic and generative AI in production at large enterprises in weeks. AWS Advanced Tier pattern partner more than doubled in a year. And they're hiring.

50 open roles. If you're someone who knows this moment is different, who wants to be inside it, not watching it, this is worth a look. At Robots and Penciles, the best ideas win, and the team is purposefully kept super high quality. This is the kind of place you look back on as the best decision you ever made. Take a look at Robots and Pencils.com slash careers. The average enterprise is spending $11.5 million on AI this year, and most of them can't prove a single dollar came back.

What does AI actually look like when it produces ROI? Ask the healthcare company that just made their payment processing 320 times faster, or the law firm whose document research went from three months to 10 minutes, or the contact center who reduced wait times by 99%. These are real Mission Cloud customers with real results. Mission Cloud is a CDW company and an AWS premier to your partner. They're the AI-first outcomes-obsessed AWS experts who build AI solutions that drive your business forward. Whether you're flooded with AI ambitions but no idea where to start or six months into a deployment that's going sideways, they've seen it and they've fixed it. Stop burning your budgets on AI that doesn't produce results. Start at missioncloud.com.

This episode of the AI Daily Brief is brought to you by OutSystems, a leading agendic systems platform built for the enterprise. Organizations all over the world are building, orchestrating, and governing agentic system on the OutSystems platform and with good reason. OutSystems open and unified platform allows teams to architect, deliver, and scale governed agenic systems with agility. Teams of any size and technical depth can use OutSystems to build, deploy, and manage AI apps and agents quickly and cost effectively without compromising reliability and security. Without systems, you can rapidly launch ideas from concept to completion.

It's the leading Agendic Systems platform that is unified, agile, and enterprise proven, allowing you to accelerate growth, reduce operational friction, and deliver real enterprise impact with AI. OutSystems. Build your agentic future. Welcome back to the AI Daily Brief. We are now officially at a week since Fable 5 was taken offline in response to new export control restrictions from the U.S. government. One of the things that that has done is cause a lot of organizations to look at their AI strategies and wonder just how solid those foundations are and how much they're relying on one, or a small handful of partners.

Part of the reason that the moment has been so resonant inside enterprise circles is that already this conversation had started in earnest, as the costs of agentic AI at the state of the art, have become clearer throughout the course of 2026. Anyone inside a big company can tell you that change is hard, and whole-scale transformation is even harder still. And there can sometimes be a tendency, especially early on in the life cycle of new technologies,

to try to reduce what matters about a change in technology paradigm, to picking the right vendor. Now, certainly some of our better-known institutions reinforce this idea, with things like the Gardner Magic Quadrant, which is all about selecting the right vendor, but when it comes to real organizational change, picking the right software is only one very small part

of the overall equation. And the lesson that we're learning in AI is that the breadth and depth with which it interacts with the existing systems inside the organization are demanding a totally different type of systems-level thinking. Now, a few days after Fable went offline, Microsoft's CEO, Satcha Dadella, took to Twitter to drop the following blog post.

It's called The Frontier Without an Ecosystem is Not Stable, and has now been seen 65 million times. It's not long, so I'm going to read the entire thing here and then get into some of the discussion around it. Satchie writes, I've been thinking a lot about the future of the firm in an AI-driven economy. This transition is different than any previous platform shift. In the past, we used digital systems to enhance human capital. This is the first time we can create a real cognitive loop between people and digital systems. That is a mindbender because it changes how we even conceptualize work inside an enterprise. What is at stake is not some digital tool or system and its use,

but how organizations continue to learn, build IP, differentiate, and thrive in a world where AI models can continuously absorb the expertise of humans and organizations and commoditize it. Every company is going to have to build what I think of as human capital and token. Human capital comprises the knowledge, judgment, relationships, ingenuity, and pattern recognition of its people, while token capital is the firm's AI capability it builds and owns. Importantly, human capital does not become less valuable as token capital grows. It only becomes more valuable.

I believe human agency will be the driver of token capital growth. Humans will set ambitious goals, connect dots across domains, build relationships, and recognize patterns that matter most. Without human direction, you have compute running in circles. This means the real opportunity is not in picking the best model, but instead in building a learning loop on top of models where human capital and token capital compound. You can offload a task or even a job, but you can never offload your learning. The future of the firm is the ability to compound that learning across people and AI.

This requires a new architectural approach, where every business is able to build agenic systems that improve over time, while still retaining control over their IP. A company should be able to switch out a generalist model without losing the company veteran expertise built into their learning system. This is the key test of your control and sovereignty in the era ahead. Companies need to turn their workflows, domain knowledge, and accumulated judgment into AI systems that improve with each use. Private evals should capture whether a model is actually improving against outcomes that matter to the business, not just external benchmarks.

Private reinforcement learning environments should let models grow stronger on real traces from inside your organization. Its knowledge base makes institutional memory queryable and use of tokens more efficient. This loop becomes the new IP of the firm. I think of it as a hill climbing machine, and unlike most assets, it compounds. Every improved workflow generates better training signal, which accelerates the accumulation of tacit knowledge unique to the firm. The companies that build this early will have an advantage that is hard to replicate, regardless of any new individual model capability. The last thing any of us want is a world where every company across every sector is seeding value to a few models that eat everything they see. If all the value is accrued by

only a few models, the political economy will simply not tolerate it. There is no societal permission for an AI future that hollows out entire industries. Think about what happened in the first phase of globalization, where entire industrial economies were hollowed out by outsourcing. The GDP numbers looked fine on the surface, but the displacement was real and the consequences are still being felt. Let us not bring that dynamic into the AI era, with a small number of AI systems capture, all the economic returns, while entire industries find their knowledge commoditized right out from underneath them. In my view, our priority has to be building a frontier ecosystem, not just a frontier model, so value flows broadly across every company, every industry, and every country,

one where every organization can own the learning loop that encodes its institutional knowledge, compounding its human and token capital. This is the ethos I've grown up with, where platforms enable more value on top than is captured inside, and where every company can continuously innovate and build value of its own. When that happens, companies will create value for themselves and the economy around them. Employees will see their expertise amplified and their judgment become parts of systems that make it replicable and scalable, and the benefits accrue to the companies and communities around them. That is how companies drive value for themselves and the broader economy, and it is the stable equilibrium we should build together. Now, one of the first reactions to this post was to see it

as a direct response to the growing power of anthropic and open AI and something of a declaration of independence, which honestly is not that hard to read given lines like, the last thing any of us want is a world where every company across every sector is seeding value to a few models that eat everything they see. It's also for those paying attention closely, clearly narrative building for the way that Microsoft seems to want their own place in the model ecosystem specifically. One thing that has not gotten nearly enough attention was Microsoft's frontier tuning announcement. Back at the beginning of the month, at Microsoft's big event, Microsoft AI CEO Mustafa Sullivan, announced the product on stage and with this post on Twitter.

It's time to move, he wrote, from renting intelligence to truly controlling your AI. Microsoft Frontier tuning lets you take our models and make them uniquely your own, turning them from capable generalists to completely custom partners. It starts with reinforcement learning environments that allow our models to learn directly from your workflows. Think of them as training gyms for AI. Here, the agent learns your very specific processes, your standards, your way of working. It goes from off-the-shelf to hyper-adaptive to exactly what you and your team's need. Those adaptations drive efficiency and performance, and your unique models can keep continually

learning in your RLE's reinforcement learning environments. This changes the nature of AI, and it changes the impact. Now, this ended up being quite prophetic to where the discussion would turn over the next couple of weeks. With one product and one idea, Microsoft is addressing a company's AI sovereignty and their AI budget all in one fell swoop. Now, folks also might point out that given how deeply enmeshed Microsoft already is in the enterprise, that this idea of an ecosystem approach, as opposed to the one model to rule them all,

eating everything approach, is in their direct self-interest as well. And yet the idea had a lot of resonance beyond that. Mark Egenstadt on Twitter wrote, I've been thinking a lot about Nadella's token capital essay. I think the whole thing boils down to one formula. Token capital equals human capital times scaffolding times feedback loops. That multiplication sign matters. If any of these is zero, your token capital is zero.

Doesn't matter how powerful the model is. I stopped asking clients about their AI model strategy and started asking about their feedback loops instead. Most companies I talked to have the model. They're paying for Claude GPT copilot. The token access is there. But their scaffolding is zero. No delivery framework, no agent orchestration, no harness engineering that teaches agents their code base, just raw prompts into raw models. And their feedback loops are zero. No cost per commit or measurement of what the AI actually produced versus what made it into production. They can't tell you whether the AI is helping or just generating noise that someone else has to clean up. Zero times anything is zero. Sightbringer writes, Sautier is describing the new

balance sheet of the firm. The old firm owned people, processes, software, customer relationships, brand, data, and IP. The new firm will own a compounding cognition loop. Every workflow becomes a training surface. Every decision becomes a trace. Every expert judgment becomes reusable signal. Every internal correction becomes model improvement. Every model run becomes a chance to turn human judgment into institutional intelligence. That is why, what token capital really means. It is accumulated, machine-operable cognition. A company's expertise becomes executable, queryable, evaluable, improvable, and portable across models. And certainly one of the big things that people are jumping on with this is this question of how to capture all

the learning that comes from AI usage. Heaton Shah writes, Satu's post is worth reading closely because it gets at the real AI question for companies. Who captures the learning? His argument is that companies are becoming a new kind of learning system. People bring judgment, taste, relationships, context, and ambition. AI brings scale, memory, reasoning, and execution. The value comes from building a loop where the company gets smarter every time work happens. The important asset is the learning system around the model. That system is built from the record of how work actually gets done. Workflow traces show the path people take, corrections reveal judgment, accepted outputs show what good looks like. Rejected approaches

sharpen the standard. Private evaluations, domain-specific context, and institutional memory give the learning structure. Over time, the company starts to retain more of what used to disappear inside meetings at its comments, decisions, and individual experience. That is the learning loop Satcha is pointing at. The judgment that once lived in a few people's heads can become part of how the company operates. Now, one immediate implication of this is for companies to want a better way to capture all that learning and accumulated experience. One of the big revelations for many of 2026 is how important to AI performance the harness not just the model is. Now, when we're talking about AI and we refer to harnesses, we're talking about the specific software layer through which we use AI that can do

things like embed context, give access to skills and tools, and generally make the AI more performant because of what you put around it. In some ways, then, what Sotcha is talking about is the need for a larger institutional AI harness that can sit around not just a model, but the entire ecosystem of AI usage within an organization. Now, there is a broader implication for companies that are really willing to go all in, which is that in many ways, the real implication is to redesign your company as a learning system that can amplify this new way of working from the ground up. And whatever the implications, Aaron Levy from Box, thinks we're about to see a whole new market response to exactly these sort of needs. He writes, the past couple months, we may be witnessing

what the applied AI layer will look like at scale. Despite some of the initial critique that this would just be a thin layer on the LLM, it's turning out that actually driving agentic workflows in an enterprise is far more complex. And anywhere there's complexity, you generally gain a mode and value over time. He then goes up. on to list some of the components of the playbook of the applied AI layer. One is building features that bridge the gap between intelligence and workflow. Aaron writes, some workflows can be automated by simply going to a general purpose interface, but others need tuned interfaces and features tied to the work they're augmenting or automating. They need features that are specific to capturing the kind of data that's

needed as context for the agent, and they need a variety of bespoke tools for the agent to use, and unique interfaces for the human-in-the-loop U-UX. Going far deeper than just presenting the output tokens, is clearly critical, and the more depth there is here, definitionally, the more sustaining value. Next, Aaron talks about how the applied AI layer can act as the model router, balancing frontier intelligence with cheaper models. This is, of course, a major theme right now and why router companies are getting so much attention. Another part of the applied AI layer is the implementation and change management, with Aaron pointing to the rise of FTEs as the example. Gabe Perrier from Harvey wrote, Sacha's article about the future of the firm really resonated with me. He describes a future

where one, every company owns their own frontier AI. Two, company's frontier AI compounds human capital instead of replacing it. Three, the compounding happens through a firm's ability to build human agent systems that learn together. Four, this results in unique and differentiated IP that belongs to the company. And five, the end result is a frontier ecosystem where value is shared across firms, their employees, and the layers they are built on. Now, if you've been listening to the show over the past couple of weeks, you will have heard about all the interesting experiments that Harvey is running in this space. But it's clear as Gabe sinks deeper into this,

that he thinks it's going to be more even than this sort of model post-training and experimentation with routers that Harvey's been doing. Applying some of the lessons of Satchez post to law firms, he writes, The cognitive loop for law firms will be a self-improving human agent system that can efficiently complete client matters end-to-end.

Doing so will require integrating a fragmented tech stack into a single platform where human and AI associates can work together to complete a client matter and learn from each other in the process. This will require completely rethinking how firms are structured, how associates are trained, how client data is protected, and most importantly, how clients are built. In other words, everything has to change. Increasingly what people

are clear on is that there really is a totally new way to work. It's one, if we are humble, that we don't know everything about just yet. Ethan Malick wrote, we don't honestly know the best approaches to rebuilding companies around AI agents, especially in ways that expand competitive advantage and augment existing human capabilities. Practical agents are merely months old. experimentation and productive failures will be required. And I think this is one of the important takeaways, especially as we head into the token efficiency era. There is going to be, I believe, a temptation to respond to increasing token costs

with strict token spend limits and biases towards known ROI. In another post, Ethan describes why this will be tempting. He writes, We are in the most comfortable, normal technology phase of AI for enterprise. It enables productivity gains but still needs integration into workflows. stuff we've seen before, yet it is very possible that this is a waypoint, not a stable phase. And I think for anyone who's really dug in with agents, it is clear that that's the case. Now, already, the most sophisticated organizations that I talked to,

were trying to shift their thinking and their actions to designing AI systems, not just AI implementations. If any good can come from the Fablify banning, it might be to put a fine point on how important that sort of systems thinking is. Interesting thoughts as we head into a weekend, but for now that's going to do it. for the AI Daily Brief. Appreciate you listening or watching, as always. Until next time, peace.