— Glossary

fiine glossary

Four definitions to understand how fiine deploys supervised AI agents: scoping, supervision, recovery, and human control.

01

The three calms method

The three calms method is fiine's approach to AI deployment, in three steps. Scoping: before installing anything, we analyse your process and set a roadmap — what to automate, in what order, at what cost range. Supervision: we move batch by batch, and every agent action stays visible in plain language. Recovery: when an agent gets stuck, it stops at the exact point and picks up again once the issue is fixed. These three steps are what make an agent hold up over time while you keep control.

02

The fiine Salon

The fiine Salon is fiine's supervision screen. Minute by minute, it shows what your AI agents are doing, how much it costs, where they succeed and where they get stuck — in plain language, on a single screen. It's the tool that makes AI legible and steerable: you see every action, every decision and every euro spent, and you keep control.

03

stop → wait → resume recovery

The stop → wait → resume recovery is the mechanism by which a fiine agent handles an error without breaking everything. When an incident occurs — an empty field, an invalid email, an unresponsive API — the agent stops at the exact point (stop), notifies you with the precise context: step, record, field (wait), then resumes where it left off once the issue is fixed (resume). No restarting from scratch, no lost work.

04

Supervised vs. autonomous agent

An autonomous AI agent decides and acts on its own, with no human oversight. A supervised AI agent keeps the human in the loop: it proposes, you validate, and it stops to alert you when it's stuck. fiine builds supervised agents — legible, and cleanly recovered when they fail — because for an SMB the point isn't to remove control, but to make it manageable.