Data in, by connection
Stripe for payments, recurring revenue and customers; Google Analytics 4 for traffic, behaviour and conversion.
02Work / AgentBoostr
AgentBoostr, a score that tells investors which AI startups are worth a second look.
Founders connect their revenue and traffic data, the platform turns it into an Agent Score, and investors get a structured way to compare one project with the next.
Overview
AgentBoostr is an incubator platform for AI startups. A founder connects Stripe and Google Analytics 4, and the platform produces an Agent Score: a graded read on how healthy the business is, with suggestions for what to fix. High-scoring projects are put in front of investors.
It was built at Taihill Venture, a Boston-area venture firm that has backed early-stage deep-tech startups since 2017. The score grew out of the framework Taihill already used to evaluate companies by hand.
What it does
4 parts
An Agent Score
Five weighted dimensions: user value, product-market fit, technical moat, business model and execution risk. Each gets a score out of 100, and the total gets a grade.
A leaderboard
Projects ranked by score, filterable by category and stage.
A path to investors
Founder and investor portals, so a strong score leads to a conversation.
What I did
3 pieces of work
- 500
- registered users
- 20+
- projects submitted by founders
- 3–4
- months to launch the MVP
01
Shipping the MVP
I launched the AgentBoostr MVP in three to four months with two engineers, building the database, the scoring backend, the dashboards, and the founder and investor portals. It grew to 500 registered users and more than 20 projects submitted by founders.
02
Founders the pipeline could not see
Some founders took payments through private channels and could not connect Stripe or GA4 at all, so the platform had nothing to score. I co-designed a way in for them: onboarding guidance, manual data submission checked by the backend, and fixes to the scoring so their numbers could be evaluated fairly.
03
Turning a framework into a model
I adapted Taihill’s evaluation framework into a five-dimensional scoring model, drawing on research into AI startups with real Stripe revenue and Google Analytics user data. The result gave investors a structured basis for comparing projects.
Screens
a2ax.ai
