I spent the week reaching out to Customer Support specialists, CS leaders, and fellow founders to ask one straightforward question:
"Where do current AI help tools actually fall short for your team?"
The feedback was raw, practical, and made it obvious why so many teams end up turning off auto-replies after the initial setup hype wears off.
Three big patterns came up repeatedly:
1. The broken human handoff. When a bot gets stuck and passes a user to a human without full transcript context, it destroys trust fast. Forcing customers to repeat themselves is a failure of the support workflow.
2. Knowledge management is the real engine of CX. Everyone competes on hallucination rates, but AI is only as good as the content layer feeding it. Uncurated document dumps lead directly to uncurated answers.
3. Perspective beats sales pitches. Approaching CS leaders as domain experts for feedback opened more real conversations (and even a partnership call) than cold trial links ever could.
I wrote up a complete breakdown of these insights and how we are addressing them with FAQ Hub over on LinkedIn
If you work in support or customer success, I would love to hear your take: what is the single biggest point of friction with AI help tools in your day-to-day workflow?
