Help validating an idea for support software
I'm building something, and I need your help.
After years running support teams at HR tech companies, I kept seeing the same frustration: agents spending 20-30 minutes per ticket searching for answers that someone else on the team already knows.
I know what you're thinking: "Aren't there already AI tools for this?"
Yes. And they all learn from your knowledge base.
Here's the problem: knowledge bases are always outdated. The best solutions aren't in your docs—they're in closed tickets, Slack messages, and your senior agents' heads.
So I'm exploring a different approach:
AI that learns from your team's actual ticket resolutions. Not what's documented. What actually worked. How your best agents solve problems in their own words.
The concept: Agent opens ticket → AI analyzes similar past tickets → Suggests response based on what worked before → Agent edits and sends.
But I'm at the very beginning. I have mockups, a hypothesis, and a lot of questions I can't answer yet:
Is this actually more valuable than existing tools?
Would agents trust suggestions based on past tickets vs. docs?
Am I solving a real problem or just a different version of the same thing?
What am I completely missing?
I'm looking for 5-10 people who are willing to look at early mockups and tell me if I'm onto something or barking up the wrong tree.
This isn't a pitch. I genuinely need to validate whether this is worth building.
And if your reaction is "this already exists" or "nobody would use this"—tell me that too. That's exactly the feedback I need.
This resonates a lot.
You’re right that the most valuable knowledge usually lives in closed tickets and senior agents’ heads, not in docs.
One thing I’d be curious about is trust: how transparent would the AI be about why it suggests a response (e.g. similar past tickets, outcomes)?
If agents can see the reasoning, adoption might be much easier.
Happy to look at the mockups and share thoughts.
In fact, it can also come from chat records, not just work orders. Simply put, our chat records are also rich in content. In my current social software, there is an AI that automatically detects the problem in your message, and automatically calls the AI to grab the relevant content in the chat record to answer the question.
I was working in a team and requested that everyone update their tickets, ideally on a daily basis. Previously the ticketing system was being used as a to-do list. Barely any description or comments.
One team member asked why and my answer was that it's a great place to find back how something was solved (validation for your idea I believe). I was able to do just that when he asked about something 6 months later.
This is how StackOverflow works isn't it? If you start a new question they will list similar questions since you might be creating a duplicate.
This is a brilliant idea! The problem you're addressing is spot-on. Knowledge bases are notoriously outdated, and the real solutions often lie in the undocumented tribal knowledge of support teams.