Matching startups with investors is the toughest dating market in the world. Both sides are looking for each other. Neither can tell who is actually available, who is genuinely interested, or who is simply out of reach. Over six weeks in Mountain View, our job was to make that market a little less blind.
The Project
Pitchtree.ai is a Silicon Valley company with a deceptively simple mission: get good companies in front of the right investors. Not more investors — the right ones. Founders routinely send hundreds of cold emails into the void, while investors drown in pitches that were never a fit to begin with. Everybody loses time, and the best matches never happen at all.
Two things stood between Pitchtree and that mission. The investor database had grown organically and nobody could say which funds were still actively investing. And the website was a single landing page that tried to explain everything to everyone, which meant it explained nothing to anyone.
So we split the problem in half. Rebekka took the data. Denis took the visibility. Six weeks later, both halves met in the middle.
Better data
We cleaned and standardised the investor database against an agreed taxonomy, then built an automated workflow that keeps it that way. It verifies recent investor activity, separates active funds from dormant ones, retrieves their investment theses, and normalises every new entry as it arrives.
The result: only investors who are actually still writing cheques make it into the matching process.
Better visibility
We grew the website from one page to seven, each written for a specific audience — founders, accelerators, investors. Every product visual is a real interface rather than a stock photo. Content now runs through a CMS, so the team publishes without a developer.
And we launched the Investor Tracker: a public directory where every fund gets its own page, structured so that search engines and AI models can read it.
Better data, plus better visibility, equals better matches. That was the formula we kept coming back to whenever we had to decide what mattered.
Being found in the age of AI
One part of the brief turned out to be more interesting than any of us expected. For twenty years, being visible online meant one thing: ranking on Google. But founders have started asking AI models instead — which investors fund healthtech in Europe, what does this fund actually look for, how do I approach them.
An AI model can only recommend you if it can read you. So every investor profile carries the firm’s name in its title, its description, and in machine-readable structured data underneath — and each one is written as questions and answers, because that is the shape a model can quote from. It is a small shift in how you write, and a large shift in who ends up finding you.
The Experience
Working in Silicon Valley does something to your sense of pace. Feedback arrived on a Tuesday evening and was expected to be live by Thursday. Nobody asked for a document describing what you would build; they asked to see it. That rhythm was uncomfortable for about a week and then became the best part of the job.
The other lesson was subtraction. Our first instinct was to put everything on the page — every feature, every audience, every proof point. Most of the real work turned out to be deciding what to leave out, and for whom. Cutting a section you spent two days building is harder than building it.
And personally? It was six weeks in a country where everything is bigger. Not always better — but reliably bigger. We would do it again tomorrow.
Project Log
| Week 1 | Kickoff with our project provider in Mountain View. Clarified scope, audited the existing investor database and the current landing page, and agreed the taxonomy. |
| Week 2 | Started cleaning and normalising the database. On the website side, mapped what each page needed to say and to whom. |
| Week 3 | Built the automated workflow for new investor data. Began the website redesign — homepage, For Startups, For Ecosystems. |
| Week 4 | Connected the Sanity CMS and launched the Investor Tracker. First round of feedback from the project provider. |
| Week 5 | Second feedback round. Researched and wrote twenty top investor profiles. Added structured data for search engines and AI models. |
| Week 6 | Went live. Wrote the handover documentation so the team can keep publishing without us. |
Project Facts
| Client | Pitchtree.ai — Mountain View, California |
| Project | Investor Database, AI Automation & Website |
| Team | Denis Thaqi, Rebekka Kaikov |
| Duration | Six weeks — 13 July to 21 August 2026 |
| Supervision | Christian Schucan, Benjamin Emmenegger — HSLU |
© 2026 HSLU | All Rights Reserved