Anthropic filed for its IPO, Nvidia reinvented the laptop, and Meta wants to be Amazon. Are we all just waiting for the market to blink?
Hi,
A few readers wrote in with a serious founder problem:
"I want to build something."
Followed immediately by:
"I have absolutely no idea what to build."
This is surprisingly common. Somewhere between "AI will change everything" and "I have 47 open tabs," many founders find themselves wandering the desert looking for a project.
So we built a list of founder experiments:
https://aiforfounders.co/founder-experiments
Think of them as intellectual kettlebells. Tiny businesses. Strange little AI creatures. Revenue experiments. Useful tools. Possibly mistakes.
Vibe coding is a craft.
It is also an art.
It is also, occasionally, a cry for help.
The point is that nobody becomes a great builder by consuming 400 hours of YouTube videos about building. You become a builder by building weird stuff that barely works, accidentally creating something useful, and then pretending it was your plan all along.
Pick one.
Spend two hours on it.
Best case, you discover your next company.
Worst case, you create a deeply unnecessary app that tracks squirrel behavior or generates breakup texts in the style of medieval monks.
Both outcomes are acceptable.
Let's go.
This Week from the AI for Founders Podcast on Spotify
The wall keeping your team out is the same wall you built to survive — LINK
Tyler Dickerhoof of Impact Driven Leader generated $700M in business sales and spent decades leading with intensity before realizing the defense mechanism that made him successful was the same one destroying his relationships. This episode is a working manual for founders on how hidden fears leak into tone, team, and revenue.
Your AI video system is broken. The prompt was never the problem — LINK
A founder nine months deep in trial and error walks through the exact model chain, character locking, and approval sequence that turns a single reference image into a consistent AI video character. The order you stitch voice and visuals matters more than any sentence you type into a box.
Your side project graveyard is not a discipline problem. It is a plumbing problem — LINK
Mariam Hakobyan co-founded Softr on one conviction: 80% of every business app is the same repetitive plumbing nobody should rebuild from scratch. Today Softr runs eight-figure revenue with 50 people, no traditional sales team, and profitable since Series A.
They failed forward for 20 years. Then they went from 7,000 to 160,000 YouTube subscribers in five months — LINK
Justin Abrams and Mike Rispoli of Cause of a Kind built a software agency on a flat monthly model and grew it by shipping video every single day. Mike's thesis on AI moats is sharp: the magic is AI plus workflow plus domain knowledge, the combination that cannot be cloned.
Hardware engineers are designing brake discs with software older than their interns — LINK
Hugo Nordell of Encube is building a browser-based CAD collaboration platform that loads complex engineering drawings in two to three seconds on a standard laptop. His core insight for founders: generative AI works in software because software has 40 years of validation infrastructure. Hardware has none, so he is building a compiler for atoms.
Nina (trynina.co) creates SEO-optimized content that ranks on search engines and gets cited by LLMs like ChatGPT, driving the kind of organic traffic that actually converts.
Investor Corner
Are you raising? Reply here, and we'll feature your deck.
AMP — First and only American-manufactured vape cartridge.
Trinity Water — Rare, Pristine, Protected drinking water.
Human Intelligence® — The only platform that can authenticate human-made content. Using patented algorithms that analyze how content is made, not just the output.
Cora — A self-optimizing website platform that uses AI to continuously test, learn, and improve your site so it converts better without you babysitting it.
EmpathEQ — AI-powered training simulations to help healthcare learners practice complex interpersonal and communication skills.
Voyager — An AI-native Web3 ecosystem and multi-asset exchange. Trade digital assets and commodities in one seamless interface.
RoboReliance — Building autonomous robotic systems designed to handle real-world industrial work so humans don't have to stand next to dangerous, repetitive machines all day.
Tip of the Spear
Anthropic files its S-1 and the AI IPO race is on — LINK
Anthropic confidentially submitted its IPO prospectus on June 1, arriving at a $965 billion valuation on the back of a $47 billion revenue run rate. Founders who have built on Claude should be watching this closely: Anthropic going public resets the entire AI pricing and enterprise negotiation landscape.
Anthropic ships Opus 4.8 with 1,000 parallel subagents — LINK
Just 41 days after Opus 4.7, Anthropic shipped Opus 4.8 with a research preview of Dynamic Workflows, enabling Claude Code to run up to 1,000 parallel subagents per session. The practical upshot: codebase-scale migrations across hundreds of thousands of lines of code, end to end, no human babysitting required.
Nvidia's RTX Spark superchip just walked into Intel's house — LINK
Jensen Huang unveiled the RTX Spark at Computex 2026, an Arm-based chip combining Blackwell GPU and Grace CPU with up to 128GB of unified memory, targeting Dell, HP, Lenovo, Microsoft Surface, and MSI. Nvidia is no longer a data center company: it is going directly after Intel and AMD on every device you own.
Google raised $80 billion right before the AI IPO wave — LINK
Alphabet is raising $40 billion in at-market stock offerings and $30 billion in underwritten shares, effectively hoovering up investor capital right before the biggest AI IPO wave in history. Whether intentional or not, Google is making the market smaller for everyone else going public this year.
Florida sued OpenAI and named Sam Altman personally — LINK
Florida became the first US state to sue OpenAI under product liability law, naming Altman personally and alleging ChatGPT causes harm to children. This lands weeks before OpenAI's expected IPO roadshow, adding regulatory risk to a deal the market was treating as a foregone conclusion.
Robinhood built a credit card for AI agents — LINK
Robinhood now lets AI agents execute stock trades and spend on its credit card with 3% cash back, no human required. If your product doesn't have an agent-native payment and trading layer by 2027, you are building for last year's user.
Zuckerberg wants to fight Amazon in the cloud — LINK
At Meta's annual shareholder meeting, Zuckerberg said a cloud computing business is "definitely on the table" if their massive data center build-out produces excess capacity. Meta is spending up to $145 billion on AI infrastructure in 2026 and companies are already approaching them weekly asking to buy compute.
Framework: The Agentic Shift
The best product is the one the user forgets they are using. That is the entire game now.
01. Audit every step in your product where the user has to think. Open your product and walk through it as a first-time user. Every moment of friction — a decision, a form, a confirmation screen, a wait — is a candidate for agentic replacement. Make a list. Rank by how much cognitive load each step carries. That list is your agentic roadmap. Start at the top.
02. Separate the tasks your user wants done from the tasks they want to control. Not everything should be automated. Users want agents to handle the mechanical and the repetitive. They want to stay in control of the meaningful and the irreversible. Map your product flows into two columns: automate and preserve. Get this wrong and you build a product that feels like it is making decisions without permission. Get it right and you build one that feels like it reads minds.
03. Design for outcomes, not interactions. A traditional product is measured by engagement: time on screen, clicks, sessions. An agentic product is measured by outcomes delivered. Rewrite your core metric. Instead of "daily active users," ask "tasks completed without user intervention." Instead of "session length," ask "time saved per week." Your investors will ask what your retention looks like. Show them that your users cannot imagine going back.
04. Build the override before you build the agent. Every agentic feature needs a clear, obvious, one-tap way for the user to stop it, review it, or undo it. This is not a safety feature, it is a trust feature. The founders who move fastest on agentic products are the ones whose users trust the agent completely. That trust is earned through visible control, not hidden automation. Build the override first. Then build the agent.
Keep It Moving
Meta's AI training tool is logging way more than it admitted — LINK
Meta's Model Capability Initiative was sold internally as a mouse-click tracker but is logging far more, including clipboard contents, code changes, and communications from EU employees who were never enrolled. EU regulators are already circling.
China's military got Nvidia chips anyway — LINK
Publicly available procurement documents show Chinese universities with military research affiliations purchased Super Micro servers containing sanctioned Nvidia chips in 2025 and 2026. The controls are leaking and Congress is now pushing for an emergency review.
YouTube will now auto-label AI videos — LINK
YouTube is moving from creator self-disclosure to automated detection, flagging videos with significant photorealistic AI and making those labels more prominent. Founders building AI video or synthetic media products should assume platform enforcement is only going to tighten from here.
Apple is building a snatch-proof iPhone — LINK
Apple is reportedly developing a sensor-based lock that triggers when the phone detects it has been grabbed from the user's hand. A small consumer feature with large implications for device security and the broader ecosystem of apps that rely on continuous authentication.
Apple's new Siri app is coming for ChatGPT — LINK
Leaked renders of Apple's iOS 27 overhaul show a fully redesigned, standalone Siri app powered by a new on-device model. Apple is finally treating the AI assistant as a product category, not a feature, and the gap with ChatGPT is narrowing faster than anyone expected.
GitHub Copilot switched to token billing and devs are furious — LINK
Microsoft's GitHub Copilot moved from a flat monthly fee to token-based consumption billing, provoking immediate backlash from developers who say their costs are now unpredictable. The move signals that the "unlimited AI" era was always temporary and usage-based pricing is the model every platform is converging toward.
Google's Gemini Spark is useful but weirdly fragmented — LINK
Gemini Spark handles inbox summaries, local event planning, and automated daily tasks, but reviewers note it is unclear why Google shipped it as a separate product rather than folding it into the existing Gemini app. Useful, but fragmented.
Google's AI can't spell "Google" and researchers say it may never — LINK
The tokenization architecture at the core of LLMs means models process text as numerical chunks, not letters, making basic spelling a structural blind spot rather than a bug that gets patched.
OpenAI is giving its pandemic AI model to governments for free — LINK
OpenAI launched the Rosalind Biodefense program, offering GPT-Rosalind for free to governments, national labs, and health institutions for pandemic preparedness and biodefense research. Early partners include Lawrence Livermore, Johns Hopkins, and CEPI.
Amazon's AI leaderboard collapsed after employees gamed it — LINK
Amazon's internal ranking system for AI adoption was scrapped after employees inflated scores by running meaningless prompts, driving up the company's own cloud costs in the process. When your AI adoption metric is gameable, it was the wrong metric.
Corgi doubled its valuation in three weeks — LINK
The AI startup raised a $106 million follow-on from the same investor set that backed its Series B three weeks prior, at double the valuation. Reflects just how fast the AI funding environment is moving.
SpaceX warned it may dilute shareholders significantly — LINK
Filed as a disclosure in SpaceX's pre-IPO prospectus, the company acknowledged it may dilute existing shareholders significantly as it raises more capital. With a target IPO valuation of up to $2 trillion, the numbers involved are unlike anything the market has absorbed before.
Blue Origin's rocket exploded. They're trying again this year — LINK
The launchpad damage is reportedly less severe than feared and Blue Origin CEO Dave Limp confirmed another launch attempt before year end.
Strava is locking down its API before going public — LINK
Strava is cutting off free API access and moving to a paid developer tier, cleaning up its data ecosystem before going public. Every pre-IPO platform eventually decides its data is worth protecting.
Nvidia and AMD's China subsidiaries may lose chip access — LINK
US officials are considering closing the loophole that allowed Chinese-owned subsidiaries outside China to purchase AI chips freely. Would complicate Nvidia's already fragile China sales strategy at a critical revenue moment.
Paris is the most important AI city outside Silicon Valley — LINK
Europe's startup ecosystem has matured to the point where top founders are scaling domestically rather than relocating to the US. Paris is now producing serious AI companies and attracting serious capital.
AI is eliminating marketing jobs faster than new ones appear — LINK
A new Martech analysis shows AI tools are replacing entry and mid-level marketing roles faster than new positions are being created. Founders hiring for growth and marketing functions should expect the talent pool and the role definitions to look very different by 2027.
Google Ads added a mode that targets people before they search — LINK
A new targeting layer in Google Ads lets advertisers reach users who fit the profile of a buyer before they exhibit active search intent. For founders spending on paid acquisition, this is worth testing before competitors figure out the right creative playbook.
What I'm Thinking About
Q1 2026 moved faster than anyone planned and most founders still feel behind. Ryan mapped exactly why, and what to do about it. — Blog
Why the prompt was never the bottleneck, and what actually moves the needle when you build AI into your workflow. — 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.
