How Forum Atlas produces every brief.

The pipeline

1. User query is embedded via voyage-3 (1024d).
2. Top-20 candidate chunks retrieved via pgvector HNSW + sector + recency filters.
3. Cohere Rerank v3.5 narrows to top-10.
4. Multi-agent ensemble synthesizes the answer — coordinator + specialists (legal, financial, technical, geospatial) where relevant.
5. Citation Validator (Haiku-class) verifies every claim against retrieved chunk substrings.
6. On validation failure: retry once with a stricter prompt. On second failure: render “insufficient evidence”.
7. Persist the brief with append-only citation rows. Audit log records the full call chain.

Citation discipline

• Every analytical claim must anchor to a verbatim substring from a retrieved chunk.
• Whitespace is normalized; case and punctuation are preserved.
• Paraphrases that don't match the substring fail validation.
• Failed paragraphs render “insufficient evidence on [topic]” rather than fabricated text.
• Citation rows are append-only — once persisted, they don't mutate.

Source quality + recency

Each source carries an importance score (1–5, canonical-vs-supplemental) and a quality_score (1–5, reliability + signal-to-noise). Both are snapshotted at citation time. The pill UX shows quality on hover. Recency labels are bucketed: just_now (< 60 min), hours, days, weeks, months, older.

Forecast methodology

Forum Forecast outputs full probability distributions, not point estimates. The model decomposes a forecast question into driver hypotheses, retrieves evidence for each, runs Monte-Carlo simulations weighted by source quality + driver-impact estimates, and assembles a posterior distribution with 90% confidence interval. Every revision tracks which drivers moved the estimate.

For a deeper read, see Citation discipline and the pipeline documentation in our developer docs.