Changelog

AI agents now see the page a visitor is on and the three pages they viewed before it. Answers can now reflect what the visitor is looking at.

Replies from your team and assignment notices show up in the chat widget straight away again.

With proactive conversations, you can initiate a conversation before a visitor asks. Show a short greeting above your chat widget based on the page a visitor is on, how long they've been there, whether they're returning, signs they're stuck, or your own custom events.

When a visitor clicks, your AI agent asks a personalised follow-up question and follows that message's instructions for the rest of the chat.

You can start from templates, preview on desktop and mobile, see how each message performed over the last 30 days, and filter your inbox to conversations a proactive message started. Available to all workspaces.

Image attachments on live messages: In the inbox, images and files on messages that arrive while a conversation is open now show up and open right away, with no page reload.

Switching workspaces: Switching workspace from an agent, customer or conversation page no longer gets stuck reloading on the "Verifying workspace…" screen.

Shopify privacy requests handled end to end: When a Shopify shopper asks for their data, store owners now get an email with a secure download link that only owners can open. When a shopper asks to be erased, their conversations are fully deleted.

Stronger chat widget security: Each chat conversation is now protected by its own private access key. An email typed into a chat's lead or follow-up form no longer links that chat to an existing customer's profile.

AI agents reply to your customer's latest message: When a customer's newest message is a short reply such as "Thanks.", the agent now answers that message instead of an earlier one.

When answering, agents now read much more of each relevant knowledge source. Detailed guides about what your business offers, like service, route or itinerary pages, now count as useful training content, so answers are fuller and more accurate.

Quality scores now take your agent's additional context, the earlier conversation and your Q&A answers into account, so well-grounded replies are no longer marked down.

We fixed an occasional "Error Loading Conversation" message when opening a conversation. Conversation rows now have tidier spacing and colours, and resolved conversations are highlighted on hover.

Weav's API reference is now generated directly from our API. Every endpoint is documented with working request and response examples, and the downloadable OpenAPI spec now includes all endpoints.