Changelog: 2026-05-31
Laid the groundwork for Canvas and early AI-assisted planning, while polishing subscription flows and improving reliability across filings, company data, and automation tooling.
- Introduced the first foundation for Canvas, including initial creation and history experiences, plus early AI-assisted planning with a safe fallback option.
- Streamlined subscription upgrades and billing management, with clearer error messages and updated trial behavior.
- Fixed an issue where shared or hard-loaded filing links could hide the primary document.
- Improved “market cap” freshness by pulling it from the canonical company data source instead of relying on stale snapshots.
- Added durable run tracking for data backfills to improve operational visibility and troubleshooting.
- Launched a new guided, step-by-step onboarding experience for registering sources and tools, with clearer governance and versioning.
- Canvas: Shipped core building blocks to support creating and tracking Canvas items, plus initial UI for entry and history. Began integrating AI-assisted planning steps behind a controllable toggle with deterministic fallback behavior when AI isn’t available.
- Billing & upgrades: Connected upgrade calls-to-action and the billing management entry point into a consistent end-to-end flow. Improved how billing errors are presented to users. Updated production trial behavior to better match current subscription handling.
- Filings experience: Resolved a rendering issue that could cause the main filing document to disappear on first load for some shareable URLs.
- Company data used in insights: Updated market cap resolution to prefer canonical company identifiers and keep results fresher in sorting and filtering views.
- Data operations: Backfill jobs now produce durable run records with summaries and failure capture, making it easier to audit outcomes and diagnose issues.
- Automations: Added an early automation capability with scheduling and run history to support repeatable, server-managed workflows.
- Security enhancements: Tightened permissions for agent access to code hosting, ensuring read-only behavior and reducing the chance of unintended changes.