TakeUp Demand Generation
TakeUp's second product. An AI-first demand-generation platform for independent hotels, built in five months.
Head of ProductNov 2025 – Apr 2026Solo build
- 5 months
- $250K
- 3

When I joined TakeUp, they were known for helping independent hotels optimize their pricing. I was the first product manager the company ever hired. My mandate was to lead TakeUp's second product, an AI-first application that generates demand for hotel rooms when organic demand is low. I scoped it from a five-word brief and shipped the MVP in five months.
How it shipped in five months.
A five-word brief became twelve product concepts, then six tested with buyers and users, then one MVP scope. The plan on the table called for three months and two contractors. Instead, I built and shipped the MVP in five months, solo, and replaced roughly $250,000 of planned contractor spend in the process. The two decks below are the actual deliverables: the scope proposal that set the build, and the pitch I used to sell to potential design partners.
Concept test results and MVP scope
The proposal I presented to decide what to build: concept testing across six ideas, the GM research, the recommended hybrid MVP, and the in/out scope.
Loading deck
Design-partner pitch
The pitch sent to prospective hotel design partners once the MVP was live.
Loading deck
What I built, and the decisions under it.
The product turns a week of market signals into three to five marketing actions a hotel GM can approve in minutes, then runs them through a human concierge. Two calls made a one-person team viable. First, a multi-model intelligence layer that sends each task to whichever model handles it best. Second, a config-driven engine where new action types are added as data, not code, so the product could grow without a release for every change.
What it became before it wound down.
Three of six warm prospects converted, alongside three of thirty-four cold leads. Three customers were booked when TakeUp wound down in April 2026. The MVP shipped two role-separated surfaces, a GM dashboard and a Concierge console, plus token-based guest access, and a closed learning loop that fed each week's results back into the next week's recommendations.






