AI Coach
Numbers first. Narrative second.
Matchup briefs, trade analysis, and ask-your-league start from structured league context. The AI must cite available evidence or abstain; it remains budgeted and fallible, not an oracle.
Evidence
before narrative
Budgets
per workspace
Abstain
instead of guessing
What's inside
- Matchup briefs 01
- Pre-week reads name key categories, schedule pressure, and risks from the latest usable roster snapshot.
- Ask your league 02
- Natural-language questions use the league context available to that surface and show evidence references when supplied.
- Trade evaluation 03
- Structured before/after with category movement, modeled outcome shift, risk, and a provider handoff rather than an automatic verdict.
- Grounding + abstention 04
- Schema checks and grounding rules reduce invention risk. Low-evidence requests should return an explicit abstention; no system can honestly promise zero model errors.
See also
Common questions
01What model do you use?+
The current implementation routes configured OpenAI models by task tier. Model names can change; the durable contract is structured context, budget enforcement, provenance, validation, and abstention.
02Is my league data sent to the model?+
Only task context should be sent, typically roster state, recent stats, and league rules. Provider credentials are excluded. See the Privacy Policy for current processors, retention targets, and operational gates.
03Can the AI execute moves on my behalf?+
No. Pick Pilot does not write moves to your provider. Review the recommendation, open or copy the handoff, and complete any roster action on the league site yourself.
Ready to use AI Coach?
Web beta. No paid checkout; provider coverage and import timing vary by league.