The teams winning in AI aren't using wrapper products. They're using coding agents to build systems. Itero Agent Skills let any agent create personas, scenarios, scorecards, and user assignments through Itero's public API. Hand your agent hundreds of call transcripts; ten minutes later you have personas modeled on real buyers and scenarios pulled from real objections — work that took weeks.
What Agent Skills do
Agent Skills are SKILL.md files that teach a coding agent how to drive Itero. Each one wraps a specific configuration surface into a callable workflow the agent already understands.
The current set:
- itero-personas — create and edit buyer personas
- itero-scenarios — author roleplay scenarios
- itero-scorecards — generate scorecards tied to specific scenarios
- itero-users — manage user accounts and team assignments
- itero-learning-paths — assign learning paths and certifications to reps
Drop the Skills into your repo, point your agent at the SKILL.md files, and the agent knows how to talk to Itero. Claude Code, Cursor, Codex, Antigravity — any agent that reads SKILL.md format works.
Wrapper products versus agent-built systems
Platforms that gate configuration behind a UI ship a thin scenario library by default. Every additional persona or scorecard costs thirty minutes of clicking. The library stops wherever customer success ran out of time during onboarding.
Configuration is a developer task. Personas, scenarios, and scorecards are JSON — defined, version-controlled, and editable like any other artifact in a repo. The UI exists for convenience; the API has always been the load-bearing surface.
The teams that win with AI roleplay aren't the ones with the best model. They're the ones whose library is deep enough that practice maps to what reps actually face on calls. Coverage is the moat, and coverage compounds the moment configuration stops being a click-through bottleneck.
You're not locked into Itero's prompts
Platforms that ship AI features inside their UI bury the prompts. The reader uses the feature; they don't see the strategy. If the prompt is wrong for the use case, the reader has no recourse.
Agent Skills invert that. Itero supplies the API; your agent supplies the analysis strategy. Whether you extract personas by industry vertical, objection type, or buyer seniority is a prompt-engineering decision you make, not a vendor decision baked into the platform.
That control is the difference between a library built around vendor-baked assumptions and one built around what your reps actually face on calls.
A real workflow: hundreds of transcripts to a deep library
Start with the data you already have. A revenue team with modern call recording sits on hundreds of transcripts. Point Claude Code at them with one of Itero's Skills loaded, and the agent works through them in parallel.
In about ten minutes, the agent returns extracted patterns: which buyer personas keep showing up, which objections cluster, which scenarios reps lose on most. You read the output, push back, refine. The agent updates the proposed personas and scenarios in natural language, then writes them into Itero.
That same flow tightens at the micro level. Author a one-off scenario — a specific objection a single rep keeps losing on — plus the matching scorecard. The agent writes both, ties them together, and assigns the scenario.
Thirty minutes of UI work, now one minute.
Try it
Itero Agent Skills are available now for every customer with API access. The SKILL.md files install into Claude Code, Cursor, Codex, or Antigravity with one command, and the agent knows what to do from there.
Want to see it work against your call data? Request a demo.