Tech Brew Ride Home - Starship Reaches Orbit For The First Time
Episode Date: September 28, 2026Nvidia launched a safety platform to stop AI agents from escaping sandboxes, SpaceX finally got Starship into orbit, Meta launched an enterprise AI business, Instinct raised $1 billion, and Trump and ...Dario Amodei had dinner at the White House. Nvidia launches the Open Agent Safety Platform, a reference design to stop AI agents from escaping sandboxes, with OpenShell for CPUs and Sentry for Nvidia DPUs (CNBC) SpaceX launches its enormous Starship rocket into orbit for the first time, deploying 26 of the most advanced Starlink satellites to join the 11,000 in service (AP) Mark Zuckerberg unveils Meta Enterprise Platform, its "next major pillar of our business" to deploy AI tools, and appoints MongoDB CEO Chirantan Desai to run it (WSJ) AI agent startup Instinct raised a $1B Series C from Sequoia, Benchmark, and Coatue at a $10B valuation and details recent products, such as a concierge service (Reuters) AI agent startup Instinct, still in early access, is building a personal assistant that runs on its own virtual phone and computer, with a phone concierge service and a trust network linking friends' agents (SiliconANGLE) Sources: President Trump hosted Dario Amodei at a private White House dinner on Sunday, their first one-on-one meeting and an indication of thawing relations (Axios) Sources: Trump dined privately with Anthropic CEO Dario Amodei at the White House on Sunday, weeks after Amodei's call for a slowdown drew the president's ire, with a tech summit on AI set for Tuesday (The New York Times) AI agents are the ultimate aggregators; they reveal apps as a means, not an end, and offering them is tech's ultimate prize, with Meta and Microsoft well-poised (Stratechery) Subscribe to the ad-free feed.
Transcript
Discussion (0)
Welcome to the TechBrewrite home from Monday, September 28th, 2026. I'm Brian McCullough today.
InVidia launched a safety platform to stop AI agents from escaping sandboxes.
SpaceX finally got Starship into orbit.
Meta launched an enterprise AI business.
Instinct raised $1 billion, and Trump and Dario Amadai had dinner at the White House.
Here's what you miss today in the world of tech.
Every day, shareholders meet to discuss important matters about the companies you invest in.
Now you can make your voice heard too.
Vanguard Investor Choice makes it easy to.
set your proxy voting preference for your eligible Vanguard index funds. Whether you hold a Vanguard fund
directly or through another brokerage firm, all it takes is a few clicks to select your proxy voting
preference and be heard on important shareholder topics like executive pay and director elections.
Visit vanguard.com slash investor choice to learn more. It's your shares. It's your voice.
It's easy. Vanguard investors own shares of our index funds and those funds own shares of the
companies they invest in. Vanguard Marketing Corporation distributor.
I'm video is launching what it is calling the open agent safety platform, a reference design to stop AI agents from escaping sandboxes with open shell for CPUs and Sentry for Nvidia DPUs, quoting CNBC.
You can't have agents roam around and drift around the company, and so you have to find a way to contain it.
NVIDIA CEO Jensen Wong told CNBC's Squackbox on Monday.
Wang said the new platform is essentially a browser for agents, providing a containment system that only allows.
allows access to things an agent needs to do its job. An NVIDIA representative told reporters on a call
on Sunday that its platform could have prevented OpenAI's hugging face incident in July. That's when
Open AIs models escape containment, access the open internet and breached HuggingFace, which operates
an open source developer platform. Each security incident is unique, and we have to look at all
of them in details, said Justin Boytano, vice president of Enterprise AI at NVIDIA, the world's most valuable
company. From what we know, Hugging Face reported over 17,000 agents attacking their infrastructure
that went on for days and weeks. One component of the platform is called Nvidia OpenShel,
which runs on central processors and sets limits on agent capabilities. Invita also announced
Century, which monitors agents and runs on network chips, not CPUs or GPUs. Some of the software
is open source, and Nvidia is calling its platform a reference design, which means
partners are intended to build products on top of it to bring it to market.
We can't have a successful AI industry if the world doesn't think it's built or confident that it's built and deployed safely, Wong told CNBC on Monday, end quote.
And quoting Wired.
OpenShall was first announced at Nvidia's annual GTC conference in March.
It's a framework for containing agents as they carry out tasks and isolating their activity in the operating system kernel, the foundational program that has access to virtually all parts of a computer system in order to,
8 hardware and software.
Nvidia's launch materials indicate that it has AI safety and security collaborations with
dozens of other tech companies, including Anthropics, Cisco, Corweave, CrowdStrike, Dell Technologies,
Hugging Face, JPMorgan Chase, Mistral, Microsoft, and Palantir.
Nvidia says SpaceX AI is using the open agent safety platform for its cursor agents and
GROC models.
The company also says Anthropic and Nvidia are building security into cloud-managed agents.
Salesforce, Scale AI and SAP are all confirmed to be integrating open-shell.
to some degree. However, it is unclear whether OpenShel has been adopted by
NVIDIA's full list of partners or whether NVIDIA is gesturing broadly.
One notable name is missing entirely from NVIDIA's list, OpenAI. Both companies indicated
that OpenAI is a part of NVIDIA's OpenShel effort, though both declined to comment
directly on why the AI lab was excluded from the announcement.
Sentry is another way for NVIDIA customers and open source users to actually implement
security policies through OpenShel, says Jessin Boitano.
While traditional sandboxes are built for application-level isolation, people now want to run fleets of agents, which demands a collective policy across all of those agencies, he says.
Agents are very creative at finding ways to achieve the goals that they're given, Boitano says.
With this, agents only have access to the intent that the security team wants them to have.
Boitano adds that Invidia is working with both Arm and Intel to create a version of Sentry that works on the X-86 chip architecture.
Once it runs on those instructions-set architectures, it can run on any architecture, he says.
Quote.
SpaceX has launched its enormous starship rocket into orbit for the first time,
deploying 26 of the most advanced Starlink satellites ever to join the 11,000 already
in service, quoting the AP.
The spacecraft almost didn't achieve the long-desired goal after one of its engines
shut down too soon, but with everything else working well and the bad engine no longer needed,
flight controllers decided after several tense minutes to proceed as planned.
Starship is orbital. Mission control announced to cheers. Founder Elon Musk's showpiece rocket, the biggest and most powerful ever built, carried 26 of the latest and most advanced Star Lake satellites to join the 11,000 older models already providing internet service. They popped out of the spacecraft one by one, drawing more cheers from the SpaceX crowds looking on from the launch site at Starbase, Texas. It was Starship's 14th full-scale launch from Texas's southern tip in three years.
The earlier test flights ventured no further than the Indian Ocean halfway around the world,
often crashing in flames and briefly skimming space.
This time, SpaceX was intent on circling the globe from an altitude of 170 miles or 275 kilometers,
not just once, but six times over almost 10 hours, ending with a Pacific splashdown near Chile.
The first stage booster wasn't meant to return to the Starbase launch site either,
instead dropping into the Gulf of Mexico within minutes of the morning liftoff, end quote.
And quoting TechCrunch.
Prior to the launch, SpaceX had planned to have the Starship upper stage orbit the Earth six times over nine hours while deploying the 26 third-generation Starlinked into the satellite company's constellation of Internet satellites.
The ship stage dispensed all of those V3 Starlink satellites during the Monday flight and was able to connect to all 26.
They are supposed to start servicing Starlink customers in the next few weeks.
This was the third test of Starship B3, each happening around two months apart.
It was also the second Starship launch since SpaceX went public in the largest IPO ever.
The company's stock jumped more than 1% in early trading Monday, but fell into the red after SpaceX decided to end the first orbital mission early.
Musk has said he wants Starship, the most powerful rocket ever built, to eventually fly thousands of times per year.
To pull this off, he recently inked a deal with Louisiana to build a new Star Base launch complex that dwarfs the existing one in Texas.
The company has said it plans to invest $100 billion in that new facility with construction starting next year, end quote.
Mark Zuckerberg has unveiled meta enterprise platform saying it is the, quote,
next major pillar of our business to deploy AI tools, quoting the journal.
Meta Platforms is starting a new business focused on selling artificial intelligence tools to enterprise customers
aiming to build on the momentum of its viral muse personal assistant and deliver a return on its hefty spending.
The move will bring together sales of several of meta's existing AI tools, including its Muse personal
productivity agent, a coding tool, access to its top models and agents for businesses, the company said.
Meta hired Chiratan C.J. Desai, previously chief executive of Document Database Company MongoDB,
to run the new business line reporting directly to CEO Mark Zuckerberg.
The Facebook and Instagram parent is in the throes of an evolution from a consumer-oriented business with a core focus on selling ads to one that also offers subscriptions and services to enterprise customers.
It is spending more than $100 billion on its AI infrastructure just this year and investors are eager to see that investment bear fruit.
Zuckerberg said in a statement that the new business line would use our strengths that few other companies have, advanced models, leading agents, large-scale infrastructure, and years of working closely with many businesses.
The company didn't provide further details on how the business would operate or price services.
Meta is entering a competitive market for business customers.
Tools such as Anthropics Cloud Code, OpenAIs, Codex, and SpaceX's cursor,
have devout followings among enterprise users and large labs face new competitive challenges
from a crop of lower-cost Chinese open-weight models.
Incumbents such as Google and Microsoft have long offered suites of workplace communication
and productivity tools as well as cloud services.
Business agents on WhatsApp, Instagram, and Messenger services announced by META earlier this summer
answer customer questions, book appointments, and close sales, among other functions.
The company has said it aims for the tools to eventually do market research and connect with tools to manage businesses' calendars.
In August, Meta also launched a coding model that AI chief Alexander Wang said was a cheaper alternative to those offered by competitors.
It is priced roughly on par with lower-cost Chinese models.
Zuckerberg earlier this year said Meta might eventually sell
computing power directly to large customers, a move that would further extend the company's business
beyond consumers, end quote.
Introducing the Pod 6 from 8th sleep. What makes it different from the Pod 5? Think of this as an
engine upgrade. Pod 6 is 50% smaller than the previous generation, packs nine times more
sensors and delivers more powerful, active heating and cooling. The pod can now read more of what your
body is actually doing overnight and feeds a better signal into autopilot.
New sizes include twin and twin XL and members see up to 34% more deep sleep.
Use code ride home at 8Sleep.com slash tech brew ride home for a special offer.
The pod six starts at $1,99 and you can have 30 nights to try it at home.
Return it if you don't love it.
Teams that struggle to scale aren't falling behind because they're lacking headcount.
or budget, it's because critical knowledge isn't documented. That's what today's sponsor,
Scribe, was built to fix. When specialists or experts leave your team, their knowledge disappears
too. That's where Scribe comes in. Scribe is a specialized intelligence platform trusted by 94% of
the Fortune 500. It captures workflows in real time and automatically generates the documentation
for how your company works with no manual writing. It automatically redacts sensitive information
like names and account numbers from screenshots,
plus your team can launch real-time on-screen guidance
showing exactly where to click, step-by-step,
inside the actual tool.
And these captured workflows
give your AI agents real context
so they can operate on how work actually happens, not guesses.
Learn more at scribe.com.
And mention techbrew ride home
for your first month of scribe capture free on select plans.
That's s-c-r-I-be-how slash ride home.
AI agent startup instinct has raised a $1 billion Series C from Sequoia Benchmark and Code 2 at a $10 billion valuation and spoke of new products such as a concierge service, quoting Silicon Angel.
Sequoia Capital, Benchmark Capital, and Co2 joined the Series C round. The funding comes a month after the company said it raised $250 million, co-led by Index Ventures and Benchmark, valuing the company at $2.5 billion at the time.
Still in Early Access, Instinct officially Spear Street Technology, Inc.
is developing a personal AI assistant that can understand the nuances of consumers' everyday digital lives.
This includes planning a cross-country road trip, ordering weekly groceries, or canceling forgotten subscriptions.
The AI agent can do this autonomously with only a few words typed on a PC or spoken into a phone.
It uses its own phone capabilities and computer to complete tasks.
We're building instinct to be the best personal agent that can.
can handle the deeply personal nuances of everyday life, said founder Noah Shin.
Over the past few months, instinct has built what the company says is a capability that lets
the agent complete phone concierge tasks.
This is a white glove service meant for handling high-touch cases involving phone calls,
high-end booking, and more such as restaurant reservations, doctors, appointments,
cancellations, or sorting out bills.
The assistant can also coordinate plans on a user's behalf with the instinct assistance of...
The assistant can also coordinate plans on a user's behalf with the instinct assistance of trusted friends and coworkers.
Using a trust network, the agents can exchange direct file sharing PDFs, images, and spreadsheets in order to quickly organize schedules,
create rapid plans, and work with a broad group to quickly sort out preferences and needs without everyone stepping on each other's toes.
This is supported by enhanced location sharing via iMessage, allowing the agent to recognize when users arrive somewhere and provide location-based personal suggestions and preference-based
responses. The company said Instinct can use devices the same way humans do, suggesting that the
core model works on a computer use paradigm. It is designed with its own virtual phone and computer,
and it uses them like a human assistant by loading webpages and apps and interacting with them
directly. Instinct is building amidst a market of roughly a dozen prominent AI agent products
and platforms now competing to become personal digital assistance. Its closest startup
rivals number only a handful, including Wadjo, Oli,
and poke, while the broader field includes agentic versions of open AIs, chat GPT, SpaceXAIs, GROC,
and Apple's Siri. Meta Platform's Muse has also built an open-claw-like personal agent and
is currently testing a similar human concierge voice call capability, end quote.
Have President Trump and Dario Amadai made up?
Quoting the times.
President Trump had dinner on Sunday night with Dario Amadai.
die, the chief executive of Anthropic, who sparked the ire of the president earlier this month when he called
for a slowdown of the rapidly developing technology. Mr. Trump hosted Dr. Amodai at the White House,
according to a U.S. official and a person familiar with the matter who spoke on the condition of anonymity.
The dinner comes as the Trump administration appears increasingly focused on meeting with AI and technology
executives amid concerns about the potential risks of artificial intelligence.
Mr. Trump, on Thursday, dined with other AI executives during a state dinner for President
Xi Jinping of China. He is also scheduled to hold a summit on Tuesday with technology executives
to discuss AI and to unveil a new government website designed to help Americans navigate
federal information, according to a White House official speaking on the condition of anonymity.
Dr. Amadai co-founded Anthropic, which makes the Claude Chatbot, and has warned about
the safety risks of AI for years. Earlier this month, he urged global cooperation in creating
rules around AI development. Mr. Trump has rejected such concerns. The Trump administration has
stopped AI people from doing bad or potentially bad things like Dario Anthropic, who is now
pretending to be a perfect little angel, Mr. Trump said in a social media post earlier this month,
and we will continue to do so. While Mr. Trump has dismissed potential serious threats around
AI as a hoax, some in the Trump administration have pushed for a more cautious approach to
the technology. Mr. Trump's dinner with Dr. Amadai may suggest an openness to hearing out those warnings,
end quote.
theory well in
Tretecory. He is making the argument that AI agents are the ultimate aggregators.
Quote, what we're seeing right now is a replay of what happened with websites.
Publications used to be scarce, dependent as they were on printing presses and delivery trucks.
Then distribution became free and the web became abundant.
That, by extension, meant the most important problem to be solved was Discovery.
The companies that solved Discovery, aggregated demand, giving them power over suppliers,
and in the case of meta and Google in particular,
incredible advertising businesses.
This is aggregation theory.
What remained scarce, however,
was actually doing things on the web or in apps.
What is happening with agents
is that the ability to do stuff is becoming abundant.
What is scarce is volition.
The problem to be solved is not discovery,
but rather inspiration.
The companies who solve inspiration
will gain power over every entity
that has things that need to be done.
Once the user is focused on solving a problem,
Every app and service required to do so is abstracted away into an implementation detail,
mere suppliers facing the fate of publications under aggregators,
scraping for crumbs from the agent, the ultimate gatekeeper of not just user demand, but desire.
I wrote years ago that the best way to understand Teams was as Microsoft's OS for SaaS.
The new co-pilot is the obvious new iteration of that.
The point has always been to own the interface and to force everyone else to integrate into that interface
on your terms. To that end, it's not a surprise that the new co-pilot isn't dissimilar to
Mews. Note that bit about autopilot including its own cloud computer, for instance, just with all
of the enterprise controls you would expect tied into it. This is the new prize in technology,
and it is the ultimate one, not just a platform like Windows or an aggregator like Facebook.
What Mews and Copilot are making a bid to be is both. The only interface you need for everything
using a computer on your behalf and generating whatever UI you need when you want it and doing
stuff without you but for you the rest of the time. What is weird about writing this all down
is that if you are actually using agents, all of this is very obvious. And yet, so many are not,
which raises the question as to whether they ever will. I think in the fullness of time,
the answer is yes. No one wants to use apps because they are apps. Apps were just the vehicle
for people to accomplish something or entertain themselves. Once people realize they can get straight
to the job to be done, it will seem odd they did it any other way. Meanwhile, definitely no one
wants to use Enterprise UI. Those inscrutable interfaces were just the way to expose arcane
functionality and lock people in. Once people realize they can simply say what they want, and that
with computer use, you don't need to depend solely on an underdeveloped and intentionally limited
API. It will seem barbaric. They did it any other way. Some enterprise companies can see this
future coming. I watched the most recent Dreamforce with interest.
is going to try and charge a three-digit premium to make their flagship product a clawed
plug-in. I salute the audacity of the cash grab in the face of computer use that will ensure
the Salesforce UI is never interacted with by a human again. What will draw the most attention,
however, is the fight to be the agent users talk to. Models are relatively substitutable.
Agents, however, operate better the more context they have about you and the more access they
have to things like your logins and files. That increases their stickiness and thus the stakes.
people and companies will only have one agent, not multiple. The challenge everyone else will face
is that the two companies first out the door with agent products that actually fit the use
case, broad exposure to a person or employee's life and environments, established messaging services,
a computer per agent, etc. are the two companies who have distribution and experience leveraging
it. Meta reaches nearly every person on Earth. Microsoft reaches nearly every employee.
The future may be distributed more rapidly than you expect, end quote.
Nothing more for you today. Talk to you tomorrow.
