Case Study - AI agent builder and simulator for complex workflows
Flote is a prototype platform that lets teams design, simulate, and iterate AI agents before deploying them into production - bridging free-form specs and working multi-agent systems.
- Client
- Flote.ai
- Year
- Service
- AI prototyping, Workflow design, Simulation
Overview
AI “agents” were the buzz of 2024, but most demos fell apart outside of a tweet thread. Startups wanted a way to design and test agents like real software, not just hope a clever prompt worked.
We built Flote, an agent builder and simulator that turned free-form feature specs into structured workflows. Teams could write acceptance criteria in plain language, then watch simulated agents plan, execute, and hand off tasks.
The platform included reusable metatags, dual markdown editors (human + agent), and views like Kanban or Gantt - making it easy to map messy requirements into repeatable processes. Instead of brittle hacks, Flote offered a safe sandbox to trial agentic behaviors before committing to production.
What we did
- AI agent simulation
- Workflow prototyping (Kanban, Gantt)
- Dual editor (human + agent)
- Reusable metatag system
Epochal's same call POC gave us a way to show what agentic systems could look like in practice. Instead of slides and promises, we had a working demo our partners could use to test real workflows.
- First working agent simulator
- 1st call
- From spec to demo
- 5 days
- Faster workflow iteration
- 3x