ShipStation Global Launches Deckhand: AI Help Desk Agent on Amazon Bedrock AgentCore
ShipStation Global's Deckhand, an AI IT help desk agent for Slack built on Amazon Bedrock AgentCore, automates answers to routine IT queries, processes error-message screenshots, and files Jira tickets. In its first three months, it resolved 97.5% of internal support queries and received a 96% positive feedback rating.
What changed?
ShipStation Global's engineering team built and deployed Deckhand, an AI-powered IT help desk agent for Slack, using Amazon Bedrock AgentCore. Deckhand answers routine IT and company questions, processes error-message screenshots, and can create or update Jira Service Management tickets. Its knowledge base is curated through Amazon Bedrock, indexing internal content from Confluence, company intranet, Slack threads, and marketing pages, all stored in Amazon S3. Over three months, Deckhand resolved 97.5 percent of Slack help desk queries and received a 96 percent positive feedback score from employees.
Why does it matter to an everyday developer?
Deckhand demonstrates how an in-house AI agent can automate common IT queries and repetitive tasks directly within team communication platforms such as Slack. By integrating multiple sources into a unified knowledge base and mining validated support threads, it delivers accurate responses and reduces manual IT effort. Its workflow shows practical approaches to curating knowledge, integrating tools (like Jira), and applying filters to maintain answer quality, which are relevant patterns for developers seeking to reduce operational friction or build similar agents.
What can the developer do now?
Developers interested in building comparable solutions can study Deckhand's architecture, which combines Amazon Bedrock knowledge bases, Strands Agents SDK, and Slack integrations. The approach includes mining and validating support threads with recursive LLM pipelines, strict filtering for answer quality, and automatic ticket filing with contextual user data. Practical next steps include exploring Amazon Bedrock AgentCore, integrating an organizational knowledge base, and implementing Slack bot workflows for automated support. The public post details prerequisites and source components for building this type of agent-driven IT automation.