Tags: New feature · UI · PerformanceWhat’s new
- Knowledge — the cf0 knowledge surface gets ten dedicated sub-views: compiled truth, drift, provenance, graph, dashboards, and more.
- Self-serve onboarding — new analysts get a guided welcome step that sets role, sectors, and language before the first Lab session.
- Help sheet, always on — a persistent help drawer with shortcut hints surfaces from any page; press
?to toggle. - Page hints — short overlays explain what each surface does the first time you land on it.
- Faster listings — Filings, Reports, and Documents lists now use HTTP caching (weak ETag + 304) so revisiting a list is near-instant. Large filings tables are virtualised — scrolling thousands of rows stays smooth.
- Dark-mode polish — token fixes for borders and muted text resolve a handful of contrast issues in dark theme.
Tags: New featureWhat’s new
- Public documentation — every cf0 surface is now documented at docs.cf0.ai. Lab, Reports, Filings, Documents, Skills, Knowledge, and analyst walkthroughs.
- Search the docs — keyword search across every page; results include the section and surrounding context.
- AI-native ingestion — every page is exposed as
/llms.txtand/llms-full.txt; the “open in Claude / ChatGPT / Cursor” dropdown copies any page straight into your AI tool. - MCP server for docs — point your Claude Code or Cursor at the cf0-docs MCP to let agents search and read the docs directly during research sessions.
Tags: Admin & complianceWhat’s new
- Public status page — status.cf0.ai goes live. Live component health for Lab, Reports, Filings, and authentication, with a feed of past incidents and root-cause notes.
- Subscribe to updates — analysts and ops contacts can subscribe to email or Slack notifications for incidents on the components their work depends on.
Tags: New feature · Data updateWhat’s new
- Excel and PowerPoint export — generate native
.xlsxand.pptxfrom any analysis or uploaded document. Formulas, formatting, and source links carry through. - Autonomous agents — overnight briefs on names you’ve analysed, ready before your morning open.
- Fund filings — N-PORT, N-CSR, and N-PX now queryable alongside 10-K, 10-Q, 8-K, and 13-F. Use these for institutional-holdings work and fund-level disclosure tracking.
Tags: New feature · Admin & complianceWhat’s new
- Org dashboard — admins see how the team is using cf0: who’s active, which skills they use, who might need help. Aggregate counts only — thread bodies stay private to each analyst.
- Team document sharing — upload a model or research file and it appears in every analyst’s workspace. Analysts can request shares; admins approve with one click.
- Shared documents power Lab — anything your team shares becomes part of cf0’s working index. Ask about a shared doc and the agent knows where to find it.
- Report version history — scrub through every version of an edited report; restore any prior state.
- Edit any uploaded document — same report-editing workflow now applies to PDFs and Word docs.
Tags: New feature · UIWhat’s new
- Report editing v1 — natural-language edits to generated reports with inline diff review and accept/reject controls per section.
/compactslash command — trigger chat compaction on demand to keep context focused.- Interrupt button — stop the agent mid-response without losing the thread.
- Unified document editing — extends the report-edit pipeline to PDF, DOCX, TXT, MD, and HTML uploads. Inline diffs, accept or reject changes.
Tags: New feature · UIWhat’s new
- Per-chat artifacts — every Lab thread keeps its reports, audits, and long-action workflows in an artifact panel (preview or PDF mode).
- Side-by-side report editing — after generating a report,
@the file in chat and ask for edits; changes appear inline with accept-or-decline controls per section. - Richer LLM output — better visualization of tables, sources, and structured data inline in chat.
- Frontend stability pass: fewer transient UI errors on long sessions and mid-stream refreshes.
Tags: New feature · AccuracyWhat’s new
- Interactive observability panel — see exactly what the model is reading from filings, financials, and sources — not just which tools it called.
- Clarifying-question gates — the model pauses to confirm the assumptions that change outputs significantly (WACC, growth rates, terminal multiples) before generating a DCF or valuation.
- Filings ingestion page — view 10-K, 10-Q, 8-K, 13-F, and S-1 filings and ingest them on demand for new companies.
- Longer chat context — long research threads stay coherent; conversations no longer drop content via early compaction.
- Speed and accuracy across queries — broad performance pass on the chat and report paths.
Tags: Data update · Accuracy · PerformanceData & filings
- Global equities — filings now extend to public equities outside the US.
- Smarter intelligence — better reasoning on multi-step questions; fewer drop-offs mid-answer.
- Accuracy improvements — tighter grounding to source documents reduces wrong figures on long conversations.
- Faster queries — chat and report generation are noticeably quicker.
- Quieter deploys — code updates ship without interrupting in-flight chat sessions.
Tags: Accuracy · Data update · UIAccuracy & methodology
- Calculation accuracy — fewer wrong numbers, fewer hallucinated figures, better cross-checking against source filings.
- SEC data speed — filings load and stream noticeably faster across 10-K, 10-Q, and 8-K queries.
- Document upload — drag-drop documents directly into Lab; sidebar shows uploaded files per chat.
- UI smoothness — chat ordering fixes so streamed thinking and tool calls render in true chronological order.
- Infrastructure stability — broad infra reliability pass across auth, sandbox lifecycle, storage, and deploy hooks.