Think-Reason-Learn Launches FounderBrain, Free Meeting Feedback for Founders

FounderBrain is a free tool that gives founders feedback on how they speak in meetings, from the transcript of a pitch, an investor call, a sales call or an interview. It launches today for AI assistants that support the Model Context Protocol (MCP), the open standard that lets them use outside tools,¹ including Claude, ChatGPT and Codex. FounderBrain is the first pre-trained model from Think-Reason-Learn, an open-source library that Vela Research, the research arm of the venture firm Vela Partners,² develops in collaboration with the University of Oxford.³Think-Reason-Learn turns a language model's judgement into questions, rules or policies that people can read and check, and runs them cheaply enough to use every day. FounderBrain comes pre-trained and ready to use, so the only thing you bring is the transcript of your own meeting.

You add a transcript and say which speaker is you. FounderBrain returns your speaking type, where you sit on four speaking styles and the well-known founders who speak most like you, as shown in Figure 1, and your AI assistant turns those results into feedback for your next meeting. Each analysis describes how you spoke in that one conversation, so you can run FounderBrain after every important meeting, or make it a weekly habit, and get feedback specific to each conversation.

Key takeaways

• FounderBrain is the first pre-trained model from Think-Reason-Learn, the open-source library from Vela Research and the University of Oxford.
• From a meeting transcript you already have, FounderBrain shows how you come across, such as how enthusiastic or certain you sound and whether you explain with data or stories, and which well-known founders speak most like you.
• Run FounderBrain after every important meeting, or once a week, for feedback specific to each conversation.
• FounderBrain is free, needs no account and keeps no transcripts.

Why is feedback on how you speak hard to get?

Founders rehearse what they say in a pitch, but few hear honestly how they came across. In a meeting, the people listening notice whether a founder backs claims with figures or with stories, and on whether they hedge or speak with conviction, yet that impression rarely reaches the founder. FounderBrain gives founders that view from the transcripts that meeting tools such as Zoom and Google Meet and note-takers such as Granola and Otter already produce.

What does FounderBrain show you?

• Your speaking type. One of five speaking types found across the 773 comparison interviews, such as The Open Enthusiast or The Candid Advocate.
• Four styles.Enthusiasm (measured to enthusiastic), openness (personal to reserved), certainty (qualified to firm) and whether you explain with data or stories, each given as a share of the comparison set, for example “more enthusiastic than 76% of founders”.
• Who you sound like. The well-known founders and CEOs of your speaking type whose four styles are closest to yours, and the well-known name least like you.
• Habits that stand out. Habits you show much more or much less often than most founders, such as putting dates on milestones or qualifying your claims.
• Three takeaways. Two habits to try and one to keep, which your AI assistant explains with examples from your transcript.

How does FounderBrain work?

The three steps from a transcript to a report usually take under a minute.

1. A small language model, trained on public interviews, answers 100 fixed yes-or-no questions about what you said, such as whether you back a point with a specific figure, like revenue or growth, and whether you tell a story from your own experience.
2. A statistical model from psychometrics, the field that measures traits such as personality, turns those answers into the four styles.
3. FounderBrain compares your styles with 773 public interviews of founders, CEOs and operators.⁴

Asking readable questions and combining the answers with a simple statistical model is the approach behind every Think-Reason-Learn method, applied here to the way people speak. Each habit FounderBrain reports comes straight from one of those questions, so when a piece of feedback seems wrong, you can check the habit behind it against your own transcript and decide for yourself, which a single score from an opaque model would not allow.

How do I add FounderBrain?

Every app connects to FounderBrain with the same address, https://mcp.thinkreasonlearn.com/mcp, and Table 1 shows where to add it in the apps we have tested.

What should I ask my AI assistant?

In a new chat, check that FounderBrain is switched on, then add a transcript and ask in your own words. Your assistant asks which speaker is you when there are several.

• “Here's the transcript of my seed pitch yesterday. Run it through FounderBrain and tell me how I came across.”
• “I'm Speaker 1 in this call with an investor. Use FounderBrain to tell me what to work on before my next pitch.”
• “I have a Series A meeting tomorrow. Run this last pitch through FounderBrain. What should I keep doing?”

What happens to my transcript?

Your AI assistant sends the transcript to FounderBrain's server, where it is analysed in memory and deleted when the analysis finishes. FounderBrain does not store or train on transcripts, names or reports. Its service logs, which never hold transcript text or names, record the type, timing, word count and any error of each request for 30 days. The company behind your AI assistant, such as Anthropic or OpenAI, keeps your conversation, including the transcript, under its own terms. FounderBrain's privacy policy gives the full details.⁵

Is there a paper behind FounderBrain?

The model behind FounderBrain is described in a paper accepted at the TAE (Trust-AI-Eval) workshop at NeurIPS 2026, the annual machine-learning conference,⁶ with the full paper to follow. Feedback helps most at this stage, so if you try FounderBrain, write to support@thinkreasonlearn.com with what worked and what didn't.