What’s New
Extended Conversation Memory
The AI assistant now retains up to 25–40 turns of conversation history, up from 3. This means entity names, segment references, and prior search results stay available across long conversations without needing to be restated. Complex research sessions that previously required repetition now flow naturally from start to finish.
Reliable Clarification Flows
When your assistant asks a clarifying question to resolve an ambiguous request, it now correctly reconstructs both the original request and the clarification exchange on subsequent turns. Previously, the assistant would lose track of what it was trying to accomplish after asking for clarification; that’s fixed.
LinkedIn Profile Lookup Stability
LinkedIn profile lookups now remain reliable even after API credential rotation. Previously, the service would hold stale credentials after a key rotation until the next restart, causing silent lookup failures. Credential updates now take effect immediately.
What This Means for You
For Sales Teams
Longer research sessions with your assistant now feel natural. Ask follow-up questions without having to repeat context, and get accurate enrichment data on prospects even after infrastructure updates.
For Marketing Teams
Complex audience research conversations are more fluid. Your assistant stays in context through multi-turn segmentation and competitive analysis without losing track of your goals.
For Operations
LinkedIn enrichment is now resilient to credential rotation. No more silent failures when API keys are refreshed.
Technical Highlights
Memory management: Conversation state now persists with a 25–40 turn window, carried forward durably across chat turns.
Clarification flow: The assistant’s internal request reconstruction correctly handles both the original query and any elicitation exchanges, preventing context loss.
Credential rotation: LinkedIn lookup credentials are validated and refreshed on every request, not cached until restart.