TherapyRecap

AI assistant for psychologists to automate your daily work.

Productivity

Artificial Intelligence

Health

I’m Sergio, a computer engineer from Spain, and I’ve spent the last year and a half building TherapyRecap. It started over a coffee in April 2025. José Antonio, a psychologist, was telling me he still wrote up his session notes by hand at the end of every day, while AI was already taking paperwork off people’s plates in almost every other field. We spent the next few months working out what a tool like this should do, and more importantly what it should never do when it’s handling therapy sessions. Then I started building. Here’s what it does today. You record or upload a session, and by the time it’s over the progress note is already drafted (SOAP, DAP, BIRP, whatever format you use). Everything you have on a patient ends up in one place: sessions, notes, test scores, old PDFs, even handwritten notes. You can ask the assistant something like “when did she first mention the insomnia?” and it answers with a link to the exact moment in the session it came from. There are a few things I’ve been stubborn about. The AI doesn’t diagnose and doesn’t score tests. It doesn’t touch the treatment plan on its own; it suggests changes and the therapist approves them or not. Patient names are swapped out before any text reaches the model, the original audio is deleted once the transcript is saved, the data stays in Europe and none of it is used to train models. We shipped 2.0 this week, and it’s the biggest change since I started. The chat used to just answer questions. Now it actually does the work: it sorts a patient’s documents into folders and puts them on a timeline using the date written on each one, drafts reports section by section with every fact linked to its source, and after each session it suggests updates to the treatment plan that you can approve, edit or throw away. It tells you what it’s reading while it works, and if you correct it once (“Marcos is her brother, not her partner”), it remembers that for good. A little over 100 psychologists have tried it so far. It’s only in Spanish for now, because I’d rather get one country right first. If you’ve built something with LLMs in a sensitive field, I’d really like to know how you handled trust. And if you’re a therapist, or know one, I’d love to hear what would make you trust a tool like this with your sessions, or why you wouldn’t. I’ll be around all day to answer anything.