FilterHN

AI Ticket Formatting Tool to help devs not waste time

6 points

22 days ago

| 4 comments

I work on my company's internal resources software (our intranet) as a dev. Employees within the company enter help tickets when something goes wrong. These tickets are often extremely vague and require follow-up. It seems no matter how meticulously we setup a form to eliminate this follow up, the tickets are still vague enough to require follow up in every case.

I just want to be able to receive a ticket and then have a really good idea of how to solve it. To do this, I'm thinking of building an AI chat bot that will eliminate as much follow up as possible so that when an employee makes a really vague ticket, the ai chat bot responds by asking them to provide specific steps to replicate and etc.

At the end of the day I get a cleanly formatted and specific ticket automatically posted to Jira that I can work on.

Feel free to roast/destroy this idea reddit-style, but if you think it has potential, I'd love to know that too! Perhaps some of you are already doing this.

▲

thirtywatt

21 days ago

[-]

I've been working on a version of this for Slack -> Jira. I built a Slack bot, Skipper, to turn conversations and unorganized messages into formatted tickets.

Generally, team members fill out mandatory fields in a Slack workflow. This gets posted in a channel and discussed. Then we @ the bot in the thread and it organizes everything in a new ticket. The AI seems smart enough to recognize mandatory fields, even when implicit.

If it helps, try it out: https://www.tryskipper.ai/

▲

skydhash

22 days ago

[-]

Wouldn't it better served by having ticket templates or forms? They can be cumbersome, but they do work. I know that LLMs can help with a more rich interactions, but I believe that the above should make for a more than good enough solution. As for me, I'd much prefer having to deal with a fixed form with a learnable layout than a step by step variable interaction with a bot.

▲

Stuart246

22 days ago

[-]

You’re essentially building a quality control layer for ticket submissions, and that could save a lot of time down the road. Definitely worth exploring, especially if your team spends a significant amount of time on follow-ups. Would love to hear updates if you decide to move forward!

▲

MattGaiser

22 days ago

[-]

How much of it is something going wrong and how much is user error? If the latter, AI could eliminate a lot of tickets from ever reaching a human.

▲

prestondavid

22 days ago

[-]

Very few tickets are due to user error, but that's likely because our PM is filtering out the ones that are user error so that I only see the ones that are problems. But you're saying that AI could eliminate the user error part?