Remember the evolution
Store meaningful changes in versioned history. Replay the state of the world at a past moment and compare it to now.
PERSISTENT, EVIDENCE-DRIVEN INTELLIGENCE
Don't just tell me
what happened. Tell me what changed.
Foresignal is designed to remember situations, test competing explanations, quantify uncertainty, and measure whether its forecasts were right.
ILLUSTRATIVE INTELLIGENCE CYCLE
OBSERVATIONS
EVIDENCE
HYPOTHESES
FORECASTS
EVIDENCE
ACCOUNTABILITYBASE RATES → SCORECARD
THINK ACROSS TIME, NOT JUST FEEDS
From raw signals to
defensible understanding.
Correlate independent evidence
Build cross-domain findings while accounting for duplicated reporting, missing observations and incomplete coverage.
Keep multiple explanations alive
Track competing hypotheses with explicit indicators and a revision trail showing why judgments moved.
Make forecasts accountable
Log predictions at a known cutoff, resolve outcomes against observed data, and score calibration against base rates.
OPTIONAL AI, NOT MANDATORY AI
Let models explain.
Let evidence decide.
Statistical detectors can identify and quantify changes. An optional hosted or local language model adds natural-language answers, briefings, and context. The same system still works without an LLM.
- Source-backed answers and citations
- Model-independent collection and triggers
- Forecasts scored like any other forecaster
- Local inference remains an option
INTELLIGENCE WORKFLOWVERSIONED
OBSERVE
New signals ingested with provenance.
CORRELATE
Connections evaluated across sources.
ASSESS
Alternative explanations tracked over time.
FORECAST
Predictions recorded, resolved and scored.
INTELLIGENCE THAT LEARNS FROM ITS RESULTS
Less noise. More accountable insight.
Explore the headless architecture behind it.