Conversations: AI chat in the support widget
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 39.9k
- Forks
- 3.4k
- Avg merge
- 6h 51m
- Merged PRs (30d)
- 232
Description
Is your feature request related to a problem?
When no support agent is online, or before one picks up a conversation, customers in the Conversations support widget wait for a human reply even for questions that could be answered instantly from docs or previous tickets.
Describe the solution you'd like
An AI chat option in the Conversations support widget: an assistant that answers customers directly in the widget, grounded in the team's documentation and knowledge, and hands off to a human agent when it can't help or when the customer asks for a person.
This pairs well with a working-hours away message (#70074): outside working hours the AI can keep helping while the widget sets expectations about when humans reply.
Describe alternatives you've considered
Agent-side AI assistance (suggested replies for humans) improves reply quality but does not reduce time to first answer when nobody is online.
Additional context
- Requested by a customer through a support ticket, together with the working-hours away message in #70074.
- The backend already has a grounded support reply pipeline (see
run_support_replyinproducts/conversations/README.md) that could power customer-facing answers in the widget.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with products/conversations/README.md and the run_support_reply pipeline described there. Trace how it could serve answers in the Conversations support widget, including grounding in documentation and previous tickets. Done means customers receive AI replies when appropriate, can request a human, and are handed off when the assistant cannot help.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- ai, backend-api-design, frontend
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100