Who IRIS Is For

What this means for you

IRIS is for the engineer who already runs more than one agent in production and is tired of pasting prompts into a notebook to remember which version shipped.

The pitch

You are past the demo. You have a real workflow with two or three agents, a reviewer, and a failure mode you want to catch before it reaches a user. IRIS gives that workflow a shape.

Who it’s for

The senior engineer or tech lead at a small AI team — the person who would otherwise be writing a YAML loader and a hash function from scratch this quarter.

Proof points

• Built for teams running more than one agent — the schema makes a single-agent setup feel like overkill
• Python 3.11+ and TypeScript SDKs ship together — covers the two languages most agent teams already use
• Fourteen CLI commands shipped at v0.1.0 — every action a tech lead needs is keyboard-driven, not dashboard-driven

quadrantChart
  title IRIS vs. adjacent tools
  x-axis "Single agent" --> "Multi-agent"
  y-axis "One-off script" --> "Reproducible system"
  quadrant-1 "IRIS"
  quadrant-2 "Notebook prototypes"
  quadrant-3 "Chat playgrounds"
  quadrant-4 "Workflow runners"
  "Notebook": [0.2, 0.2]
  "Single-agent SDK": [0.25, 0.55]
  "Workflow runner": [0.6, 0.4]
  "IRIS": [0.85, 0.9]