Changelog - June 26, 2026
See exactly what context your agent had
Every AI response is now backed by a context snapshot you can inspect. Open the Show context drawer on any AI message in a conversation to see the visibility-tagged system prompt segments and the inputs the agent used to generate that reply — so when you're reviewing a conversation, there's no guesswork about why the agent said what it did.
- Per-message transparency: Each AI generation captures the exact context that produced it.
- Segmented system prompt: The system prompt is broken into labeled segments so you can see what the agent was told.
- Review with confidence: Diagnose unexpected answers by looking at the real inputs, not a reconstruction.
Relevant docs:
Keep your agent available after a visitor engages
A new Show globally after engagement setting lets you override your visibility rules once a visitor is in a conversation. Turn it on and, after a visitor starts or resumes a chat, the agent stays available across your entire site — including pages that don't match your visibility rules — for the rest of that visit and when they return later.
- Engagement-aware visibility: Visibility rules still gate the first impression, but an engaged visitor keeps the agent everywhere.
- Persists across pages and visits: The agent follows an engaged visitor as they browse and when they come back.
- Opt-in: The setting appears in your agent's customization when you have visibility rules configured.
Relevant docs:
Reference anything in your prompt with @-mentions
The agent builder's prompt editor now supports @-mentions. Type @ and reference tools, questions, and captured fields directly inside your instructions. Mentions render as pills in the editor and resolve when the prompt is built, so your instructions stay accurate even as the underlying configuration changes.
- Mention your configuration: Pull tools, questions, and fields into the prompt by name.
- Always in sync: References resolve at prompt-build time, so renaming or editing the underlying item keeps your prompt correct.
- Readable pills: Mentions appear as clean, grouped pills instead of brittle hand-typed names.
Relevant docs:
Rep profile photos
Teammates can now upload a profile photo. Your reps' photos appear across the product and in the live chat experience, giving visitors a more human handoff when a conversation reaches a person.
- Upload a photo: Add a profile photo from your user settings.
- Shown where it matters: Rep photos surface in the app and in live chat.
Connect Aimdoc to your AI tools: remote MCP server
Aimdoc now exposes a remote MCP (Model Context Protocol) server, so external AI clients and assistants can securely connect to your Aimdoc data and act on it. Access is protected by OAuth 2.1, with dynamic client registration and token-based authorization — no static API keys to copy around.
- Standards-based access: Connect MCP-compatible AI clients to Aimdoc over a streamable HTTP endpoint.
- OAuth 2.1 security: Clients register and authorize through a standard OAuth 2.1 flow with scoped, revocable tokens.
- Built for the agentic era: Bring your Aimdoc conversations, visitors, and leads into the AI tools your team already uses.
Relevant docs: