OpenAI got its GPT-5.6 hall pass, Anthropic's ad scared everyone, and Musk swore he won't pull the plug. Trust, but verify?
Hi,
a weird thing has happened.
I accidentally became a deal flow pipeline.
I talk to dozens of founders every week. Some are raising. Some should be. Some probably shouldn't be, but confidence remains undefeated.
Either way, I keep meeting exceptional founders building companies long before they show up with LinkedIn victory laps, or someone else's portfolio page.
so… an experiment.
If you're an angel, family office, or VC actively looking for great founders, raise your hand. I'm putting together a shortlist of investors who want quality introductions from the front lines.
no fees. no platform. no AI matching your aura.
Just thoughtful intros between good founders and investors who are actively writing checks.
If that sounds useful, hit reply and tell me what you invest in, your typical check size, stage, sectors, and geography. The next great deal might be on my calendar.
Let’s go.
This week from the AI for Founders podcast on Spotify :
Your website is gaslighting you — LINK
Eric Schneider of Cora built a bootstrapped website optimization platform that added one client $20K a week in revenue and projects a 24x lift in yearly conversion by rewriting the site per visitor, and his sharpest founder take is that AI should stop imitating humans and start acting like AI.
The soccer league nobody controls — LINK
Tal Melenboim, the serial entrepreneur behind Movota, Score:Plug, and VFR.ai, built Lega.bot, an autonomous 24/7 AI soccer universe with thousands of independent agents, and his rule for founders is that your biggest unsolved problem is hidden value your competitors haven't found yet.
Snipd is the world’s best podcast streaming app // try premium for free as a sub from this newsletter!
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
Personalized menswear styling subscription
Less shopping, more living - dress for success with Taelor → Save $30 with RYAN30
Tip of the spear:
Apple sued OpenAI over alleged trade secret theft — LINK
Apple accuses OpenAI's hardware chief Tang Tan and other ex-Apple staff of using confidential project code names and coaching recruits to evade security. If proven, OpenAI's hardware ambitions were built on stolen blueprints.
Satya Nadella says you're paying for AI twice — LINK
Nadella argues enterprises pay in tokens and then hand over the "exhaust" of their corrections, quietly training the next competitor. His fix: build your own learning loop instead of renting someone else's.
Demis Hassabis calls for a FINRA-style AI watchdog — LINK
Hassabis wants an industry-funded, government-backed standards body to test frontier models before release. Self-regulation with teeth, or the industry writing its own rulebook first?
GPT-5.6 cleared for broad rollout after government review — LINK
OpenAI's three-tier model family (Sol, Terra, Luna) got the green light after weeks of delay. Reviewers still say Anthropic's Fable has the edge in raw intelligence.
New York becomes the first state to freeze new data centers — LINK
Gov. Hochul's year-long moratorium targets anything over 50 megawatts, citing utility bills and grid strain. 71% of Americans oppose data centers in their backyard, per Gallup.
200+ economists admit they're "driving in fog" on AI's economy — LINK
Sixteen Nobel laureates and the chief economists of OpenAI and Anthropic signed a joint statement admitting the profession can't yet measure AI's real labor impact. The people building the tools don't know what it does either.
Samsung's chip division will out-earn 40 years of history in one — LINK
Samsung's 2026 semiconductor profit is projected to beat its cumulative earnings since 1985, driven by AI memory demand. SK Hynix separately warns the shortage won't peak until 2027.
The real AI race may no longer be at the frontier — LINK
Hugging Face's Clem Delangue says Chinese open-weight models now outdownload US frontier labs, and most production workloads may end up running on cheaper open alternatives. The moat everyone assumed existed might already be gone.
Most plant-based athletes are under fueled.
Momentous is the only brand that meets every standard expectation.
Without compromise. My favorite protein and creatine here at 20% disco for readers.
Framework:
The Three-Tier Filtration System
Every growing founder hits the same wall: too many inbound requests, not enough hours, and no system to separate the urgent from the noise. This week's framework treats your inbox like a factory line, not a to-do list.
01 — Build the trash filter first. Anything that doesn't require your unique judgment (scheduling, FAQs, intros) gets auto-routed to an AI assistant or a template reply before it ever reaches your eyes. If a bot can answer it in one sentence, a bot should.
02 — Create the batch lane. Everything that needs your attention but not your immediate reaction (partnership pitches, press asks, non-urgent requests) gets bucketed into a single weekly review block. Touch it once, not five times as it resurfaces.
03 — Reserve a same-day lane for anything tied to revenue or reputation. Customer escalations, investor questions, anything with a dollar sign attached jumps the queue automatically. Build the rule into your triage system, don't rely on remembering to prioritize it.
04 — Audit the lanes monthly. What was "same-day" six months ago is probably "batch" now that you've hired for it. Filtration systems rot if you don't retune them as the team and tools change.
05 — Protect the founder layer. The entire point of tiers one through three is to buy back your attention for the 5% of decisions only you can make. If you're still personally triaging tier-one noise, the system failed, not the flow of inbound.
Agentic partnership acquisition for the world’s top podcasts. —> gethowdi.com
Keep it moving
OpenAI ships GPT-Live-1, a voice model that can talk and listen at once — LINK
The new full-duplex model replaces Advanced Voice Mode and can hold 30-40 minute conversations. Live translation demo still had a heavy accent problem.
Meta's Muse Image lets strangers remix your public Instagram photos — LINK
Anyone can tag a public account and use those photos in an AI-generated image, with no notification to the person whose face it is. Opt-out exists, but it's buried in settings.
Meta launches Muse Spark 1.1 to compete with Anthropic and OpenAI on coding — LINK
Zuckerberg posted on X for the first time in three years to promote it. Pricing undercuts Claude Haiku 4.5 and GPT-5.6 Luna by a hair.
Elon Musk praises Anthropic, promises not to cut off its compute — LINK
Anthropic pays SpaceX's xAI unit $1.25 billion a month for Colossus compute through 2029. Musk says he was "clearly wrong" calling them losers last September.
Phia accused of cookie-stuffing its way to affiliate commissions — LINK
Phoebe Gates and Sophia Kianni's shopping startup allegedly overrode other affiliates' referral codes mid-checkout. Impact.com suspended the company pending review.
CISA admits it had no incident playbook during its own breach — LINK
A contractor leaked AWS credentials to a public GitHub repo; CISA staff built its response plan mid-crisis. The agency has been without a permanent director since January 2025.
SpaceX cleared to fly Starship again after May's booster failure — LINK
This is the first Starship flight since SpaceX's IPO. It'll carry the first third-generation Starlink satellites.
NHTSA issues an ultimatum to robotaxi makers over first responders — LINK
The agency wants "solutions" by month's end after Waymo vehicles repeatedly interfered with emergency scenes. San Francisco is separately investigating a July 4th gridlock caused by stranded robotaxis.
A humanoid robot assisted in a live gallbladder surgery — LINK
UC San Diego researchers used a remotely operated Unitree G1 for seven procedures. Every movement was human-directed; nothing was autonomous.
OpenAI kills its Atlas browser after eight months — LINK
Its features move into a Chrome sidebar extension instead. Atlas joins a growing pile of scrapped OpenAI products, including the ChatGPT Agent and Sora.
Apple's foldable iPhone may launch with almost no supply — LINK
Analyst Ming-Chi Kuo expects only 500K-1M units in Q3, with delivery delays of 4-6 weeks. Scalpers could flip units for a 50-100% premium.
iOS 27's public beta arrives, and the new Siri is actually good — LINK
9to5Mac calls it the most stable iOS beta in years. Full rollout lands this fall.
Apple signs $30B+ deal with Broadcom for US-made wireless chips — LINK
The deal covers 15 billion chips and expands a Colorado plant. It's part of Apple's $600B US investment pledge under tariff pressure.
DeepSeek is reportedly building its own AI inference chip — LINK
The move would reduce reliance on Nvidia and Huawei amid export controls. OpenAI and Anthropic are said to be doing the same.
Meta locks down Ray-Ban smart glasses to stop privacy-light tampering — LINK
A mandatory update disables the camera if the LED is drilled out or covered. Meta is also pursuing legal action against people selling the modification.
Claude Cowork expands to mobile and web — LINK
Tasks can now run and finish with your laptop closed. Business process automation, not coding, is the dominant use case: content and copywriting alone beat software development.
ChatGPT commands 92% of AI referral traffic, but Claude is quietly climbing — LINK
Claude grew 64x since late 2024 and overtook Perplexity in March. One vendor's product tweak can still halve your AI-referred traffic overnight.
Anthropic's newest ad is creeping people out — LINK
"There's hope in hard questions" pairs a burning house and a graveyard shot with the line "who's gonna hit the brakes?" Sam Altman called it satire-adjacent on X.
Vercel's Guillermo Rauch on decoupling models from agents — LINK
Rauch says clients are done picking one lab and betting everything on it. Open models like DeepSeek and GLM-5.2 are "taking off" inside Vercel's own gateway data.
If you use Google, you're training its AI, here's how to opt out — LINK
A quiet settings change now lets Google retain uploaded images, voice search audio, and Lens photos for model training by default.
Inbox Alchemy grows your list and publishes your weekly founder newsletter.
Email is the only distribution motion you actually own . Apply Now
What I’m thinking about:
Stop chasing superintelligence, customers just want reliability — 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 .
