GaiaHunt
& ScavAgent
A city game platform used for large group hunts and team-building events. Players join through their phone browser; an AI authoring workflow helps prepare the hunts.
The product
Players open a link on their phone, follow a route, and solve clues at real checkpoints. The platform supports photo missions and team scoring, with a shared gallery and leaderboards.
Marco owns the product decisions and implementation, including payment flows, organizer tools, and production operations.
The AI workflow
The authoring pipeline moves through a blueprint, structured generation, review, and targeted repairs. Gemini supplies the model output; Langfuse records the stages so failures and costs can be examined.
Generated content remains a draft. Human curation and validation of the route on location are required before publication.
ScavAgent
The companion MCP server lets an author work on a brief inside ChatGPT or Claude. Marco built its OAuth 2.1 authorization server and the integration with GaiaHunt’s creation workflow.
Hunts in practice
GaiaHunt has been used for multiple hunts with more than 100 participants. It has also hosted team-building hunts. Marco’s work includes running these events as well as developing and maintaining the platform. Start and end times, scoring, and final leaderboards are automated.
The engineering
The stack includes Next.js, React, TypeScript, and Supabase, with Cloudflare R2 for object storage. Jest and Playwright cover automated checks, alongside field testing.
This is the practical background behind our AI rollout consultancy and SaaS and app development.
Visit GaiaHuntWhat’s your
next move?
Tell me what you want to build or improve.
A short description is enough to start.