True North: Give AI a job worth doing
Pick the right problem. Give the product room to change. Stay close enough to the people using it to know whether it helped.
Those ideas connect a lot of what we’ve been learning this month — from the Connected Stack stage to our fellows’ summer projects. Inside: software that could adapt to each customer, a tutor that asks students to explain, and why the next version of your product might need less of what you’ve already built.
Signal
You have to change how you work.
At Connected Stack, Anthropic’s Kate Jensen and True’s Mike Montano discussed what it takes for an enterprise to adopt AI: a committed business owner, a clear outcome, and the willingness to change a process. Access to technology is one part of the investment. Organizations also have to change how they work.
What if every customer got a different product?
On the Cadre AI podcast, Rohit Sharma explored what software could become if the cost of customization and maintenance approached zero. The founder’s question changes: which compromises could you remove if you could afford to serve each customer differently? Listen to Rohit with Cadre AI’s Chad Lohrli → · The LinkedIn clip →
“Would we build it this way today?”
Alan Wells of Rocketable asks a question worth putting into your next product review. As models improve, code that once made a product possible can become work you no longer need to maintain. Revisit the reason it exists before adding another layer. Alan on knowing when to rebuild →
Inside True
The conversation has moved into deployment.
Connected Stack brought founders, builders, and investors together around how companies deploy, secure, and trust AI at scale. We’re grateful to the co-hosts who curated the stage with us, the partners who made the day possible, and every founder who shared what they’re building. The Connected Stack recap →
Eleven fellows, ten companies, a summer of building.
Our eighteenth fellowship class built customer-call briefings, a searchable lab inventory, a podcast production system, and more. One project took a parenting app’s chat feature from parent interviews through prototype to implementation; another brought a fellow who had never written iOS into building a mobile CRM chatbot.
The lesson we keep coming back to: follow the friction. Some fellows realized their first version had moved the problem somewhere else, then rebuilt. Give someone a real problem, access to the people doing the work, and room to revise. Julie’s fellowship recap →
Founders Building
Aristotle
A tutor that asks you to explain.
Aristotle’s voice tutor shares a whiteboard with students, remembers where they got stuck, and asks them to explain their reasoning. Shan Reddy, Jaiden Reddy, and Vivek Vajipey have deep roots at True — Shan invested here, and Jaiden and Vivek were True Fellows. We led their seed round, and we’re excited to see what they build for students and families. Why we invested → · Meet Aristotle on LinkedIn →
TokenCompass
Who’s doing the work — and what are they spending?
Ian Chan’s TokenCompass brings AI spending across providers into one view, broken out by person, agent, team, model, and project. Customers can see what agents spend or roll that spending up to the people who deployed them. As more work goes to agents, leaders need a clearer way to see where the money goes. Why we invested → · Mike on TokenCompass →
ICEYE
From Erasmus students to a €10 billion valuation.
ICEYE founders Rafal Modrzewski and Pekka Laurila met as students. Today, their company is valued at over €10 billion.
This month, European Commission President Ursula von der Leyen highlighted ICEYE in her State of the Union address. Its radar satellites help customers see through clouds and darkness, with uses spanning emergency response, insurance, and environmental monitoring. ICEYE’s latest milestone →
Polsia
Give powerful tools a specific job.
Polsia founder Ben Broca’s starting point was helping people start businesses. His advice to founders: choose a clear customer and a narrow job, and explain the product in terms that customer understands. He also reminds us how much room remains to help people adopt AI beyond Silicon Valley. Ben on finding your first customer → · On the adoption opportunity →
Sailplane
The hardware has to get to work.
Sailplane’s Sam Ramji points to the expertise required to get complex AI hardware running and keep it reliable. He also makes the economic case for faster deployment: capital sits tied up while a rack waits to produce useful compute. Time to deployment deserves a place alongside performance in the conversation. Sam on hardware complexity → · And why deployment speed matters →
Team Time
The opportunity after the code is written.
In a conversation with MTS, Puneet Agarwal predicts the market for deploying AI-written code could become larger than the market for generating it. Testing, security, governance, and tracing all become more consequential as companies produce more software. Watch Puneet on what comes next →
Why bring competing venture firms into the same room?
Tony Conrad and Madeline Minshew joined MTS ahead of Connected Stack to talk about collaboration and what it takes for companies to put AI to work. Listen to the conversation →
Madeline is now a Partner at True.
She leads our Founder Experience through an AI-native lens, building the programs, tools, and connections that accelerate our founders as they build legendary companies. We’re so happy to celebrate this next chapter with her. Read the announcement →