An internal Slack agent resolved most routine IT questions
In brief
According to AWS and ShipStation Global, ShipStation built Deckhand, a Slack AI agent that answers routine IT questions from a knowledge base, analyzes error screenshots, and creates or updates Jira tickets. Over three months it resolved 97.5 percent of questions in that channel and received 96 percent positive staff feedback. In the year before launch, employees posted more than 1,500 threads, over a fifth outside US working hours.
What it changes for an SME
This is a custom build on Amazon Bedrock AgentCore, not a product you can buy off the shelf. If staff already repeat the same IT questions in Slack and answers already live in Confluence or intranet pages, a similar assistant can reduce repetitive work for IT and cover requests outside working hours. Recreating this stack (hosted agent, knowledge index, ticket tools, screenshot analysis) is a real engineering and running cost, which a company of 10 to 250 people should compare with actual ticket volume. The article does not cover cost, GDPR, or the EU AI Act.
What you can do
- Count how many repetitive IT questions staff send in chat or email each month, and whether the answers already exist in your documentation.
- Check that those documents are accurate and who is allowed to see them before any assistant indexes them.
- If volume is high, decide whether you also need automatic ticket creation in your current help desk, not only answers.
Let's talk about the time you could save.
In 20 minutes, we find where AI could save you time or money. Free, no commitment.