Movement
Summaries can explain what changed in market-implied probability and which public signals contributed.
AI Forecasting Research
Veritas uses AI to explain market movement, not to erase uncertainty.
AI summaries in Veritas are a research layer over public market signals. They describe probability movement, narrative pressure, freshness, and uncertainty without advice language or guaranteed outcomes.
Probability movementCalibration watchSignal evaluationResearch only
Public identityVeritas Prediction Intelligence Engine
Veritas is a prediction-market intelligence and forecasting engine for analyzing event markets, probability movement, market signals, and research-grade forecasting workflows.
Free layerPublic market intelligence
Private betaAPI / MCP workflows
BoundaryNo advice or guarantees
AI Interpretation Layer
Useful summaries need source context and explicit limits.
Narrative heat
AI can group visible movement by category or topic while avoiding unsupported certainty claims.
Freshness
Summaries must reflect live, delayed, cached, or snapshot data states instead of hiding feed quality.
Uncertainty
The language keeps confidence, sample size, volatility, and resolution timing visible.
Governance
Private prompts, tool permissions, exports, and MCP actions remain scoped and auditable.
Risk boundary
AI output is research context, not financial advice, betting advice, or a guaranteed forecast.
Summary Contract
What a safe AI forecast summary should disclose.
The summary should name the market context, describe the observed probability movement, identify freshness state, and state the uncertainty boundary. It should not convert descriptive analytics into instructions to trade, bet, or allocate capital.
Context
Market, category, source
Movement
Probability and volume change
State
Live, delayed, cached, snapshot
Limit
No advice or guarantees
AI Boundary
What the public AI layer is not allowed to imply.
No guaranteed prediction accuracy
No trade instruction
No betting recommendation
No private prompt exposure
No hidden model claim
No jurisdiction-free availability claim