👍 AI Assistant — Share Feedback — CCG Platform Engineers
Date: September 28, 2026 Audience: Internal CCG Platform Engineers Status: ✅ Live
Every answer from the CCG AI Assistant chat panel now has a quick thumbs up / thumbs down control, with an optional comment on a thumbs-down. This is the fastest way to flag a wrong, outdated, or unhelpful answer directly to the team building the Assistant.
Same audience as the AI Assistant itself: CCG Platform Engineers, not exposed on the merchant-facing site.
What's New?
Each assistant answer in the chat panel now shows "Was this helpful?" with thumbs up/down buttons. Voting thumbs-down opens an optional "What went wrong?" comment box before submitting.
| Capability | Detail |
|---|---|
| One-click rating | Thumbs up/down on the latest assistant answer in a conversation |
| Optional comment | Thumbs-down prompts for an optional note on what went wrong |
| Confirmation | "Thanks for the feedback!" on success; a retry prompt if submission fails |
| Triage-ready | Feedback is stored with the question, answer, and citations so the team can trace exactly what was wrong |
| Alerts on signal | A thumbs-down or a comment notifies the team; a lone thumbs-up doesn't page anyone |
How It Works
- Ask the Assistant a question as usual in the chat panel.
- Rate the answer — click 👍 or 👎 under the response.
- On 👎, optionally explain what went wrong — or click "Skip" to submit without a comment.
- Feedback is tied to that conversation turn automatically; you don't need to copy/paste the question or answer.
Rating controls appear on the most recent assistant message in a conversation — this is intentional, so feedback always maps to a specific, unambiguous answer.
Why Are We Doing This?
| Reason | Detail |
|---|---|
| Fastest signal we have | A thumbs-down is far lower-friction than filing a ticket, so we expect to hear about real misses sooner |
| Traceable, not just a number | Feedback is stored with the question, the answer, and its citations — enough to actually debug a bad answer, not just count votes |
| Feeds future eval work | Down-voted questions are a natural source of new test cases for the Assistant's evaluation suite |
What You Need To Do Now
- Nothing required — just use it. Rate any answer that's wrong, outdated, or unhelpful.
- A comment on a thumbs-down is optional but genuinely useful — a line like "cited the wrong doc" or "answer is for v1, I asked about v2" goes a long way in triage.
📚 Resources
| Resource | Link |
|---|---|
| AI Assistant (Chat) — original announcement | AI Assistant (Chat) |
| Documentation Home | docs.healthsafepay.com |
🏁 Bottom Line
You can now rate any AI Assistant answer directly in the chat panel. Use it — a quick thumbs-down (with or without a comment) is the most direct way to flag a bad answer.
Having trouble with the Assistant? Use the feedback link on any page or contact your integration support representative.