02 / Case study
MiroxAI SaaS · Own product
A multi-agent AI platform for Amazon ads, built end to end.
Sellers lose hours tuning PPC bids by hand. In Mirox, specialist AI agents analyse the campaigns and recommend every bid, with the reasoning shown.
16specialist AI agents orchestrated in production
01Problem
Amazon sellers spend hours analysing PPC campaigns and adjusting bids, keywords and budgets by hand, and the automated tools that exist are black boxes they don't trust with their ad spend.
02Solution
A SaaS platform where 16 task-specific Gemini agents analyse advertising data, forecast conversion and recommend bids, each decision explained. Simulation Mode runs for 14 days on the seller's own account before anything goes live.
03My role
Founder and lead developer. I designed and built the architecture, the Next.js dashboard and marketing site, the FastAPI backend, the agent orchestration server, the database and the deployment.
- Product idea
- Architecture
- Frontend
- Backend
- Database
- AI agents
- Infrastructure
- Production
04Complexity
Keeping 16 agents coordinated and every decision explainable, and earning enough trust that sellers hand real ad spend to an AI.
05Result
A production SaaS that onboards sellers, analyses their campaigns and produces transparent bid recommendations. Its sister product, optimirox.com, was built the same way, end to end.