Improvements
- Gain finer control over analytics with a new setting to track opens for all emails (instead of sampling).
- Sequence generation is smarter: only relevant activities are considered and reasoning messages are clearer, making Autopilot behavior easier to understand.
- Duplicate “Answer question” tasks with the same due date are now combined into a single, clearer item.
- Calendar tasks are no longer created when all external participants have declined an event, reducing noise.
- The context panel now shows a loading state while internal team members are being enriched, so you know when data is on the way.
- Saved filters are more reliable and scoped to your workspace; empty filters are automatically cleared for cleaner persistence.
- Import apps now feature refreshed icons (Zapier, Vector, Segment) for a more polished, consistent look.
- Import qualification now shows clearer “unqualified” reasons, improving review decisions.
- Cal.com imports display the organizer’s email when a name isn’t available, improving event details.
- Action names in contact activities are clearer: “Schedule reply for date” and “Schedule follow-ups.”
- Activity messages and prompts better prioritize your instructions, clarify nurturing rules, and note that contextual knowledge comes from your knowledge base; unarchive reasoning now includes the campaign name.
- Tooltips can show either an info or question icon based on context, improving in‑app guidance.
Fixes
- Campaign sequence generation now correctly uses 1 credit, ensuring accurate credit usage and billing.
- Archived contacts with a pending auto‑archive task no longer appear in the activity log.
- Filters saved in your browser are now correctly scoped to your organization and persist more reliably across sessions.
Behind the scenes
- Introduced action sync infrastructure with event‑driven handlers and scheduled jobs to keep tasks up to date.
- Optimized contact summaries by recalculating conversation history only after a reasoning run and removing redundant cron events.
- Hardened contact processing: skip contacts with recent suggested imports, delay processing for brand‑new contacts, and only enqueue processing when needed.
- Improved Autopilot suggestion handling by setting processing states more reliably, both in background tasks and direct database updates.
- Added A/B testing for action determination to refine prompt strategies by organization and environment.
- Expanded import reliability with a logs array for better traceability, plus refined schemas and validation (including a simplified Cal.com schema and organizer email fallback).
- Added a comprehensive open tracking mode on the backend and updated forms and documentation to support it.
- Refined sequence generation internals with stricter activity filtering for cleaner variables and reasoning.
- Simplified navigation history code in the app shell to reduce complexity and improve maintainability.
- Enhanced import qualification internals with additional fields to track unqualified reasons more precisely.