🚪Enter the dungeon of feedback hell
Recently, at work, we had an AI hackathon event, and I had to come up with an idea for an AI tool.
So I cheated, I created a tool of something that I'd already done for myself, I created an AI feedback crawler.
In my spare time, and make a few simple websites, I write and host specialist quiz nights, and write and host murder mystery nights.
I love getting feedback from all these things; the more feedback, the better. So I made a simple feedback form to be used for any event or website.
But, there was a problem. If I'd run a week of quiz nights, I might have had 1,000 people attend and received 600 feedback messages. I loved the feedback, but couldn't handle it all. So I built a feedback engine, first on a personal level, and then for the aforementioned hackathon.
⚔️ Meet Feedback Crawler Carl
Feedback Crawler Carl is an AI-powered feedback engine that does what I don’t have time to do.
Instead of just collecting feedback, Carl acts on it.
- Praise? → Carl writes a proper response (not the usual “thanks 👍”)
- Bugs? → Carl creates GitHub issues (and can even start fixing them)
- Feature requests? → Carl groups them and pushes them into Jira
- Trends? → Carl spots patterns across everything
Basically… Carl is the party member you wish you had in every sprint.
🧩 The Real Problem (That Most Teams Ignore)
Where does the feedback come from in software companies, usually from a decision maker, the person who decides if someone uses your software.
That does make great business sense, but does it make sense for the development of your software?
Why not get feedback directly from the end users, all of them.
Feedback Crawler Carl is simply a feedback button that sits on every screen of all your applications.
The users can then send feedback, good or bad, and it all goes to the same place, to Carl.
🧠 What Carl Actually Does
🟢 Praise Handling
Carl doesn’t just say thanks — he:
- Personalises responses
- Spots trends across products
- Suggests improvements based on what users love
🔴 Bug Handling
Carl goes full developer mode:
- Creates GitHub issues automatically
- Can kick off fixes via Copilot
- Learns from recurring bugs (knowledge base style)
🔵 Feature Requests
Carl:
- Groups similar requests
- Identifies high-impact ideas
- Creates structured Jira tickets
🟡 Insights & Trends
- Detects sentiment shifts
- Surfaces patterns across feedback
- Helps prioritise what actually matters
🔁 Closing the Loop
This is the bit most teams forget, Carl tells users when things are fixed.
Which means:
- Users feel heard
- Decision-makers get nudged to upgrade
- Everyone wins
🏆 The Results
Personally, for my websites and events, this AI-powered feedback engine has helped me focus on the things people think are important, rather than focusing on what I think is important, and that has really helped.
Professionally, well, I didn't win the popular vote in the hackathon, and righty so, other ideas would be better for generating profit, but I did manage to show how AI can be put to create use in improving that often huge gap between developers and end users.
Top comments (0)