Changelog

Channels — our most-requested feature — are now available on all Pro+ plans. Connect Gmail, WhatsApp Business, or Telegram and your AI agent responds automatically.

• NewChannels (Gmail, WhatsApp, Telegram) now available on Pro plan and above
• NewAuto-send mode for high-confidence replies (configurable threshold)
• NewFreeScout native integration — replies appear directly in mailbox
• NewGorgias integration added for Shopify helpdesk teams
• ImprovedDraft reply quality improved with full conversation context
• ImprovedInbox Reply now includes source citations in the draft
• FixedFixed draft duplication when webhook fired twice on same ticket

Every conversation is now scored for sentiment and CSAT. See how your support quality changes over time — and catch frustrated customers before they churn.

• NewSentiment analysis on every conversation — frustrated, neutral, satisfied
• NewCSAT prompt at end of chat sessions with daily trend chart
• NewChurn-risk signal: flag conversations with negative sentiment trend
• NewSentiment available on Business+ plan; CSAT on Pro+
• ImprovedAnalytics dashboard redesigned — CSAT and sentiment on one view
• ImprovedSource performance chart now shows answer quality by knowledge source
• FixedCSAT score occasionally not saving for mobile widget sessions

Connect any REST endpoint to your AI agent — no code required. The AI agent decides mid-conversation when to call your tools, fetches live data, and includes it in the reply.

• NewNo-code API tool builder — connect any REST endpoint
• NewTool result formatting — define how the AI agent presents live data
• NewTool call history visible in conversation logs
• New30 tools per AI agent on Starter, 80 on Pro, 120 on Business, unlimited on Business+
• NewShopify Orders API pre-built template for WISMO queries
• ImprovedTool execution timeout increased to 10 seconds
• FixedTool auth headers now encrypted at rest

Complete rewrite of the AI agent engine with a new reasoning layer, confidence scoring, and intent detection. Faster, more accurate, and far more configurable.

• NewReasoning layer — AI agent searches, scores confidence, then responds
• NewConfidence threshold — configure when to escalate vs answer
• NewIntent detection — recognise cancel, escalate, book, return intents
• NewBrand dictionary with categories and usage tracking
• NewPrivate mode — restrict widget to authenticated users
• Newagent cloning — duplicate config to new workspace in one click
• NewRLHF training tab — mark conversations trained or irrelevant
• ImprovedAnswer accuracy up 40% vs 1.x on benchmark test set
• ImprovedWidget load time reduced by 65% (lazy bundle, edge-cached)
• RemovedLegacy keyword-matching fallback removed

• NewUnanswered questions report — see exactly what the AI agent couldn't answer
• NewSource performance analytics — which knowledge sources drive answers
• NewExport conversation logs as CSV
• ImprovedAuto re-sync now configurable — daily, weekly, or on-demand
• FixedPDF indexing failed silently for password-protected files

• NewVoice input in chat widget (Web Speech API, Starter+ plan)
• NewNotion workspace sync — pages, databases, and sub-pages
• NewGoogle Drive folder sync — Docs, Sheets, Slides
• ImprovedYouTube transcript ingestion now handles auto-generated captions
• ImprovedCrawl depth increased to 500 pages on Business+ plan
• FixedWidget Z-index conflict with certain WordPress themes

• NewAI conversation summarization — one-click summary for agent handoff (Pro+)
• NewInternal conversation notes — team annotations visible only to agents (Starter+)
• NewConversation tags — categorise sessions for reporting
• ImprovedSummary quality improved for multi-turn technical conversations
• FixedNotes occasionally shown to end users in widget — resolved