Changelog: 2026-08-16
Social post creation and review got a major upgrade with stricter evidence requirements and a guided revision workflow that produces clearer, more traceable updates.
- Strengthened evidence requirements for automated social-post flags, emphasizing freshness and direct relevance.
- Introduced a guided revision flow in Slack with structured change requests and refreshed, versioned review cards.
- Raised the bar for revisions to ensure substantive rewrites rather than light paraphrasing.
- Improved end-to-end observability for the revision cycle with clearer approve/reject outcomes and review scoring.
- Expanded market data capabilities with new macro data tools and intraday anomaly detection.
- Automated candidate selection for social posts now applies additional quality filters (including configurable thresholds) to reduce low-signal drafts and focus on more meaningful coverage.
- Drafts must include clearer, evidence-backed company context, with a tighter writing style that prioritizes falsifiable facts, quantification before interpretation, and source-backed comparisons.
- Content constraints were reinforced to avoid investment instructions, price targets, sensational claims, and other unsupported framing, keeping posts focused on verifiable information.
- Reviewers can now request changes through a required Slack modal, which triggers a regeneration workflow and publishes a new, versioned review card once ready—keeping the previously reviewed version intact until updates succeed.
- Revision requests can steer new research while being treated as editorial direction; outputs still must meet the same evidence and accuracy standards.
- Slack review cards now distinguish internal evidence links from what gets published, reducing confusion during review.
- Evidence gating for automated flags was tightened to require fresh event dates, relevant evidence anchors, and explicit citation of the trigger source in the opening claim.
- Market data tooling expanded with macro series access (tier-gated) and a new intraday microstructure layer that can score unusual volume/price behavior for anomaly detection.