RAIDT Studio
run-level evidence
🏠
Dashboard
overview
🧠 GenAI
🤖 Agentic
📚
Start Here
guide · learn
⚙️
Configure & Run
set up the run
▶ Run
📈
Evidence & Scores
totals · records
💡
Improve
recommendations
🎛️
Calibration
SME sign-off
▶▶ Run ALL
🗄️
Saved runs
experiments
⚙ Advanced ▾
🤖
Agentic pipeline
trajectory test
⬇ GenAI
Calibration
Results
Tab design
⬇ Report
⬇ Bundle
«
‹
›
»
idle
Full report
⬇ Export JSON
✕ Close
Provider
deterministic — no key, reproducible
openai — GPT (needs key)
anthropic — Claude (needs key)
groq — Llama/Mixtral (needs key)
together.ai (needs key)
openrouter (needs key)
ollama — local (no key)
Model
Stress-test repeats
Scoring method
Buckets (default)
Weighted (graded)
API key
GenAI
Agentic
Prompt for
—
· the real prompt this controller will run for the selected domain. Click a controller to load it; edit to override. Each selected controller runs separately and is compared.
Domain
Default dataset
Run ALL
Count
Start #
Auto-play
Slow
Normal
Fast
Datasets root folder
Custom dataset (optional)
Reset
Auto-walk
Activity log (live during a run — ■ Stop anytime; completed records viewable while it runs)