True North: Your failure data is gold

In this edition of True North, three things we've been learning right alongside our founders: the most valuable data is often the stuff nobody thought to keep. Building our own tools taught us more than watching ever did. And the founders who launch something every month — win or lose — learn faster than the rest of us. Also: we'll keep your seat warm at this fall's enterprise AI event if you sign up at connectedstack.ai →

Signal

Your failure data is absolutely worth keeping. We're seeing a trend: companies building their whole positions on what most of the industry throws away — failed edits, null results, samples nobody saved. It works because if you train only on public wins, you lose all the edge case learning, which is ultimately some of the most valuable data there is.

Launch every single month. Forever. One founder in our community treats every month as a chance to launch something: a feature, a press hit, a new market, a re-ship. Each launch should have one BHAG (big hairy audacious goal). Forever launching takes the pressure off any single launch and gives you regular opportunities to learn, iterate, and improve.

Some product-market-fit is the most dangerous place to be. Because it's just enough to delude yourself into thinking you have real PMF. Failure isn't what kills most companies at the earliest stages. A little traction is. Run the ceiling analysis on your best quarter, not your worst: if everything works, is it still big enough?

Inside True

Twenty years of our own write-ups, now queryable in real time. Our AI Builder in Residence Keerthi Nalabotu spent her first eight weeks at True turning all our firm's documentation into True's Datalake — one query layer over everything the firm knows. On top of it: a network atlas of everyone we know, a copilot that sits in on diligence calls, and a loop that catches its own wrong answers. It's gold. Read more from Keerthi →

The Journal covered our office hours. WSJ on how venture firms are actually using AI — including True AI Office Hours, built by our partner Mike on Polsia, a True portfolio company. Read it →

Founders Building

Ask Claude to design an antibiotic. Basecamp Research's foundation model is now accessible in Claude Science and across Anthropic's products — the deepest map of natural biology we have, sitting where scientists already work. Rohit on why this one matters →

Founder Mode Maxxing. Handshake's Garrett Lord shares how Handshake AI went from $0 to $1B in fifteen months. The beginnings of this wild ride trace back to one Christmas party conversation. Garrett tells the story →

A whisper is the interface. Sandbar's Stream ring takes notes, dictation, and real-time voice chat at a whisper — haptics confirm it heard you, a tap and swipe handles your music. See how it works →

Three more eyes on the planet. Three ICEYE satellites went up from Vandenberg. The radar doesn't care about clouds or darkness. Watch it →

A day in the life of a founder who sells sleep. Hatch's Ann Crady Weiss gets hers. Her routine, for the rest of us →

Open Tabs

Tab 1 A place on the internet for looking out strangers' windows.

Tab 2 True Fellow Pear Chotbunwong's summer, as told through an advent calendar.

Team Time

Welcoming Gretchen Robinson to True Ventures — our first Head of Investor Relations.

Om Malik was a legend. No words or video can quite capture the force of him. We'll miss you, friend.