Project Outcomes
Hometeam Ventures tracks thousands of construction and housing technology startups, with roughly 50 new companies added weekly. Every company required manual tagging across five dimensions. The process was slow, inconsistent, and dependent on one person’s expertise – limiting how quickly promising solutions could be found and shared with investors.
This project set out to design and deploy an AI classification pipeline that tags companies automatically at the point of entry, while keeping human judgment in control of quality. The system combines an off-the-shelf AI agent for immediate classification with a confidence-gated review queue, so uncertain cases route to an analyst rather than enter the database unverified. The work also required building workflows to integrate with other systems.
Hometeam Ventures gained more than an automated tool. The team now spends less time on manual tagging and more on higher-value verification and sourcing, and the firm can search its database directly – democratising access to information and accelerating how quickly capital reaches solutions addressing the global housing gap.
Key technology/tools used: Figma, Claude LLM, Dust.tt AI agents, Notion, APIs, ReactJS, Express.js, Google OAuth & Vercel.
Flowchart diagram
Homepage of tagging application
