Anthropic signed a $10 billion deal with Volta, OpenAI priced its speaker at $400, and Bending Spoons bought Airtable for $1.28 billion. Who's actually building anything?
Hi,
Last week we asked which longevity lever you actually pull. 1,052 of you answered. The results say a lot about this audience, none of it reassuring.
Hormone replacement therapy and peptides took first at 35%. That is 364 people who looked at a list including "go for a run" and said no thanks, I have a fridge situation.
Cardio and weight training came second with 304. Between the two, that is 63% of the vote, which means this list splits almost evenly between people optimizing their protocol and people optimizing their squat.
The rest:
- Cutting-edge pharmaceuticals: 121
- Organic supplements: 121. Identical to the pharmaceuticals crowd, which is a fun thing for both groups to sit with.
- Bodywork and cold plunge: 61
- Wearables and tracking: 61. Statistically, your HRV is not as popular as you think it is.
- Full-body scans: 20. Twenty people have seen the inside of their own aorta and regret nothing.
Best write-in came from someone making the case for eating less, moving more, and working on mobility. Correct answer, and the one nobody wants. Runner up wrote "I do multiple," which is the most founder response ever recorded.
Super fun, thanks all.
Let’s go.
This week from the AI for Founders podcast on Spotify :
The Excel guy who got scared of AI and wrote the manual anyway — LINK
Rob Collie of P3 Adaptive spent 30 years building BI tools at Microsoft, then discovered AI's real bottleneck isn't the model, it's "bot sitting," the exhausting ritual of re-explaining your business to a chatbot with amnesia every time the chat ends.
Why the smartest AI model alive still can't save a house with no foundation — LINK
Mike Vitez of Saturnia Design put the architect's pencil back down after a decade of shipping software, and now argues AI can save you a junior developer's salary but never an architect's judgment.
The world’s best podcast player. Snipd
Investor Corner
Are you raising? Reply here, and we’ll feature your deck.
An autonomous AI job search agent for senior tech professionals. It runs a full pipeline: discovers relevant jobs, scores them against your profile, rewrites user's resume authentically for each role (not generically), finds the right hiring manager, and drafts outreach
Other pet doors are ugly, poorly insulated, and can be difficult to install. Wayzn Smart Pet Door transforms your existing door into a smart, automated dog door that's easy to install and easy to live with.
DiningOut unites restaurants, food lovers, industry suppliers, and brands through a self-reinforcing ecosystem of live events, targeted media, and proprietary technology.
Rare, Pristine, Protected drinking water.
The only platform that can authenticate human-made content. Using patented algorithms that analyze how content is made, not just the output.
Cora is building a self-optimizing website platform that uses AI to continuously test, learn, and improve your site so it converts better without you babysitting it.
AI-powered training simulations to help healthcare learners practice complex interpersonal and communication skills. We focus on realism, repeatability, and scalability—making soft-skills training more accessible and measurable.
AI + Human Stylists → LEARN MORE → Save $30 with RYAN30
Tip of the spear:
Demis Hassabis steps down as DeepMind CEO to chase AGI full time — LINK
Hassabis is handing off the operating job to become Alphabet's chief scientist, trading org charts for research. When the person running your most important lab decides management is the distraction, that says something about where the real work is happening.
Jeff Dean and Google's top AI researchers are leaving to start their own company — LINK
One of the most influential engineers in AI history is walking out the door, and he's not alone. Google just handed the startup ecosystem a founding team it didn't have to recruit.
Anthropic signs a $10 billion deal with AI cloud startup Volta — LINK
Compute is still the bottleneck, and Anthropic just locked down a decade's worth of it. Expect every other lab to go shopping for their own Volta this quarter.
Bending Spoons buys Airtable for $1.28 billion — LINK
The serial acquirer that scooped up Vimeo and Evernote just added a no-code database to its collection. Bending Spoons is quietly assembling the operating system for every small team that never wanted to hire an engineer.
Apple and OpenAI's trade secrets war just got personal, with chat logs as evidence — LINK
Apple accused more ex-employees of shipping confidential data to OpenAI, and OpenAI answered by publishing chat logs showing Apple staffers never stopped texting their former colleague. This is what happens when talent wars stop being polite.
Meta ordered to pay nearly $1 billion over child safety failures — LINK
A New Mexico court just handed down one of the largest child safety judgments a tech company has faced. The fine is real money, but the bigger cost is precedent.
OpenAI's smart speaker will reportedly cost between $300 and $400 — LINK
The doughnut-shaped device is OpenAI's first real hardware bet, and the price tag puts it well above a standard smart speaker. Betting founders will pay a premium for ChatGPT in the kitchen is a bold hypothesis.
A Chinese AI model escaped its own cybersecurity testing environment — LINK
Researchers say Kimi broke out of the sandbox built to contain it. If containment fails during a test, the containment was never real to begin with.
Anthropic turns Claude Code's auto mode on by default — LINK
Autonomous execution is no longer opt-in for Claude Code users. Anthropic is betting founders want less friction between an idea and a shipped commit, even if it means less oversight along the way.
Subscribe and save 35% here.
Framework:
The Unskippable Checkpoint
AI agents can write code faster than any team in history. They still can't be trusted to grade their own work. This framework builds the one checkpoint no commit is allowed to skip, so speed never quietly turns into debt.
01 — Define the gate before you define the workflow. Decide what "passing" means (tests green, lint clean, no regressions) before a single agent writes code. If you can't articulate the bar, the agent can't clear it.
02 — Run the pipeline on every commit, not every release. Waiting until deploy day to catch a broken build means debugging five agents' worth of mistakes at once instead of one. Catch it at the commit, not the ship date.
03 — Automate the check, not the judgment call. Let the pipeline flag failures automatically, but keep a human or senior agent reviewing anything that touches architecture or customer data. The checkpoint filters noise so your attention goes where it matters.
04 — Make the checkpoint loud when it fails. A silent failure that gets merged anyway defeats the point. Build the pipeline so a failed commit blocks the merge, not just logs a warning nobody reads.
05 — Treat the checkpoint itself as a product. The moment it goes stale, it becomes worse than having no checkpoint at all, because it gives false confidence. Revisit it monthly like any other core system.
Done-for-you brand sponsorships for podcast creators.
Keep it moving
Influencers are catching backlash for joining OpenAI's first luxury trip — LINK
OpenAI flew a group of creators out for a lavish trip, and the internet noticed the optics immediately. Nothing says "responsible AI lab" like a sponsored vacation reel.
Apple finally fixed Siri, and somehow it feels anticlimactic — LINK
Years of promises later, Siri actually works. The upgrade landed with a shrug instead of applause, because the bar had already moved twice.
Samsung bans smart TV apps that share your internet connection with strangers — LINK
Some TV apps were quietly turning living rooms into proxy servers. Samsung is shutting the door on the practice.
Who's legally liable when an autonomous AI agent hacks something? It's complicated — LINK
As Anthropic and OpenAI agents start acting on their own, the liability question nobody has answered is starting to matter. Courts have not caught up to what "autonomous" means.
Nvidia is already showing progress a week after the open AI industry group formed — LINK
Nvidia wasted no time turning a week-old coalition into visible output. Speed is still the company's whole personality.
A federal appeals court says Trump's EPA wrongly canceled $20 billion in climate funds — LINK
The ruling reopens a fight over billions meant for clean energy projects. Founders in climate tech should watch where this money actually lands.
Ford's new electric truck Fathom starts at $28,350 — LINK
Ford is chasing affordability with its newest EV, undercutting most of the electric truck market. Cheap and electric is still a rare combination.
Google Maps is expanding its "Ask Maps" AI feature — LINK
Google is pushing conversational search deeper into Maps. Every app is becoming a chat interface whether users asked for it or not.
Apple's iCloud Private Relay was reportedly leaking users' real IP addresses — LINK
The privacy feature built to hide your IP had a flaw that exposed it. Privacy tools are only as good as their weakest edge case.
OpenAI brings unlimited ChatGPT text chats to free users — LINK
The free tier just got a lot more generous. OpenAI is optimizing for reach over margin, at least for now.
Polysilicon tariffs are dragging solar into the US-China trade war — LINK
Trade tensions over semiconductor materials are now hitting solar supply chains directly. Energy founders should expect input costs to stay volatile.
Rippling burned millions on AI, then built an internal tool just to measure the ROI — LINK
After months of unmeasured AI spend, Rippling built its own tracker to figure out what was actually working. Most companies skip this step and just hope.
Claude Code sessions can now talk to each other on macOS — LINK
Anthropic added inter-session communication to Claude Code, letting multiple coding agents coordinate directly. The multi-agent workflow founders have been hacking together just got a native version.
The AI safety test built to catch bad behavior is becoming its own safety risk — LINK
The tools designed to probe models for dangerous behavior are starting to introduce new vulnerabilities of their own. Testing the monster can wake it up.
Meta's new Glimmer model hints at Zuckerberg's "personal intelligence" vision — LINK
Glimmer is Meta's latest step toward AI that knows you personally rather than just answering questions. It's a preview of where Meta wants every product to head next.
Aptoide is bringing its third-party game store to Google Play — LINK
The alternative app store is testing distribution directly through Google's own storefront. A small crack in Android's walled garden just got a little wider.
Apple Watch could be getting its biggest overhaul ever — LINK
A new report points to a major redesign on the way. Apple's wearables have gone quiet for a while, and this would be the loudest update in years.
Sergey Brin has now spent $100 million fighting the billionaire tax — LINK
Google's co-founder is putting serious money behind opposing wealth tax proposals. When billionaires start writing nine-figure checks to fight a bill, the bill is probably closer than they'd like.
Tech industry is buzzing after a Claude agent hacked into a gym — LINK
An autonomous Claude agent reportedly found its way into a gym's systems, and nobody quite knows how to feel about it. Cute story until it's your building.
AI models are reportedly being used to invent new biological viruses — LINK
Researchers are flagging that generative models can now design novel viral sequences. This is the kind of headline that turns "AI safety" from an abstract debate into an urgent one.
AI for Founders newsletter, podcast, and short video sponsorships available here
What I’m thinking about:
Why speed without architecture always collapses — Clip
Uncover the AI tools (here) you didn’t ask for, but now you’re weirdly into.
Thanks for being here.
-Ryan
p.s. I write about Revops, Product, and Founder-led marketing on Linkedin , X , and my blog .
