Hi everyone,
We've been exploring Copilot's suggested replies feature and ran into a couple of things that have us a bit puzzled. Hoping someone here has experienced something similar or can shed some light!
For context, we use WhatsApp exclusively and have our knowledge base set up with Q&A entries.
The "email" thing
When a suggested reply gets skipped, we hovered over the ⓘ icon and saw this tooltip:
"Did not suggest a response because it looks like the original email does not contain any question."
The message was sent via WhatsApp. Why does the tooltip reference "email" here? Is this expected behaviour, or something else?
The matching feels very literal
This one was eye-opening. We have a Knowledge base article titled "Can I use contraception while breastfeeding?", Copilot only generates a suggested reply when a user types that question verbatim. The moment someone asks anything slightly different e.g., "Can I take family planning while breastfeeding?" it skips entirely, even though they're asking the same thing.
For a WhatsApp audience, that's a real limitation. People don't phrase things the same way, and it feels like the feature needs a bit more flexibility to be truly useful on a conversational channel.
A few questions for the community:
1. Has anyone found a way to improve Knowledge base matching for varied phrasing, like adding synonym-aware entries or alternative question formats?
2. Why does the skipped reply tooltip say "original email" for a WhatsApp message?
3. Is this something the paid Copilot add-on ($20/seat/month) addresses, or is the underlying matching logic the same across tiers?
Would love to hear if others have worked around this. Thanks in advance!
