If you put five developers on five different AI coding tools, chances are you’ll get five different results from the same ticket. Each team member uses a different set of prompts and has developed different habits while using a given AI agent. One person will paste the requirements into a chat window by hand. Another might neglect the acceptance criteria specified in a Confluence page. AI is certainly capable of writing something plausible, but it's often forced to guess the context that was one tab away the whole time.
Helper AI
The Importance of Shared Context
The hard part of adopting AI by an engineering team is making sure that model actually understands the work, regardless of how great it is. Every developer ultimately ends up building their own prompts, their own conventions and shortcuts, and their own way of copying a ticket into a chat window. The problem with this is inconsistent output and no reliable way to know why one developer's AI-assisted work looks different from another's.
HAI simply makes the tools your team already has work better together. It favors consistency over convenience: everyone needs to agree to a single way of working up front. The benefits are stronger: the whole team gets the same context, the same tooling, and far fewer inconsistent results to untangle later.
HAI’s Features
- HAI adapts itself to your project and the AI tools your team already uses. After setup, it provides a command to tailor the prompts and agents to your project's structure and code, avoiding the use of generic text. It configures itself inside your AI coding agent of choice and matches each tool's own configuration, so nothing has to be done by hand. Teams without a strong editor preference can also rely on HAI's general setup, which works anywhere.
- Turns tickets and specs into something AI can use. Jira issues and Confluence pages get pulled down as readable text files, comments, links, and structure included. If a ticket references other tickets, those get pulled in too, so AI obtains the big picture rather than one isolated ticket with missing context.
- Handles screenshots and attachments. Files attached to a ticket often require a login to access, which normally blocks an AI from seeing them at all. HAI fetches those attachments with the right credentials and makes them available locally.
- Splits the work efficiently. Under the hood, HAI uses a set of specialized roles: one that researches a task, one that plans it, dedicated ones that write backend, frontend, and refactoring code, and others that review the result for security and correctness. Those roles work in parallel rather than one after another, which shortens the time from ticket to working code.
- Reduces working on tickets to a single command. Once set up, a developer can hand HAI a Jira ticket number and it takes care of the rest: pulling the context, researching the codebase, drafting a plan, and, once approved, implementing the change.
Who Is HAI Intended For?
It is built for:
- Leaders standardizing how their team uses AI. Developers naturally gravitate toward different tools. HAI lets them keep their preference while making sure the underlying workflow, and the results it produces, stay consistent.
- Teams working off Jira and Confluence, where the ticket and the spec that explains it live in separate places an AI assistant can't reach on its own. HAI automatically closes this gap.
- Engineering managers who are tired of “tribal” knowledge and looking to unify AI usage. To put it more simply, people who want every developer's assistant configured the same way instead of relying on whatever each person happened to set up for themselves.
- Organizations that don't want another platform, database or dashboard, and would rather have tickets and specs as files they can read, review changes to, and commit like anything else in the repo. HAI doesn't introduce a new dashboard or database to manage.
Getting Started
Setup is a one-time task, and it's built with simplicity in mind.
1npm install -g @2amtech/hai
2hai initRunning hai init walks the team through choosing an AI provider and where tickets and specs should come from. It also runs the first sync automatically, so there's no separate setup step before the tool is usable. Based on their preferences, teams can choose to:
- Point HAI at Jira and Confluence
- at local files if they prefer to keep documentation in the repository,
- or skip ticket integration entirely and use HAI just for its shared workflow and roles.
Free, Open Source, Apache 2.0 Licensed
HAI is released under the Apache 2.0 license, which means it's free to use, modify, and build into commercial projects without restriction. Data pulled from external sources, such as tickets, specs, and credentials, remains on your machine and is excluded from Git by default. HAI does not add a proprietary service layer or route your project data through additional infrastructure.
In other words, all data stays yours and is kept locally; nothing is sent to the server. There’s no new system to manage.
If your team is already spread between different AI coding editors, or if context keeps getting lost between the ticket, the Confluence spec, and the AI doing the work, HAI is more than worth setting up.
