AI-MI
AI token creation platform- PERIOD
- 01.2025 — 06.2025
- ROLE
- System architecture · full-stack
- TEAM
- 4
- STACK
- React · NestJS · OpenAI API · GCP · Terraform · Docker
- LINK
- Sunset · 2025 — no longer live
[ screenshot — AI-MI — token launch + agent chat ]
Users launch their own tokens on EVM networks — each paired with a dedicated OpenAI-powered agent — and swap them without leaving the platform.
What I did
- Designed the overall architecture across frontend, backend and AI-integration workflows.
- Built frontend interfaces and backend APIs for the core business flows.
- Deployed and operated services on Google Cloud, with infrastructure managed in Terraform.
- Integrated on-chain token swaps with the team on EVM networks.
01 Problem
Launching a token is easy; giving it a reason to exist is not. Most new tokens have no utility beyond the chart.
02 Approach
A token factory on EVM networks where every token ships with its own OpenAI-powered agent, plus in-app swaps — backend on Google Cloud, infrastructure in Terraform.
03 Result
Shipped to production in six months; the product was sunset in 2025. The architecture and IaC patterns carried into later projects.
Infrastructure as code from day one
Every GCP resource was defined in Terraform, so environments were reproducible and reviewable — and the whole stack could be stood up or torn down cleanly.