Ideanation

Inspiration

As a developer, I’ve found that AI code generation only works well when requirements are clear, structured, and testable and most real-world requests aren’t. Clients and stakeholders often describe what they want vaguely, leaving developers to infer intent, edge cases, and acceptance criteria.

I’ve been manually using ChatGPT to spec features before coding, but the friction was obvious: specs live in Confluence and Jira, not in a chat window. I wanted a way to turn raw domain knowledge into high-fidelity specs where teams already work, and to make those specs reliable inputs for AI-assisted development, verification, and QA.

What it does

Ideanation is a Rovo-powered app that turns vague problem descriptions into spec-driven, AI-ready delivery artifacts directly inside Confluence and Jira.

In Confluence, users can describe a problem in natural language and use Rovo actions to generate a structured problem statement, personas, a Day 1 persona, challenges, risks, ideas, and user stories. Those stories can be pushed directly into Jira.

In Jira, Rovo expands each story into clear functional and non-functional requirements and generates BDD scenarios, giving teams testable, verifiable specs that are ready for engineering, QA, and AI code generation tools.

How I built it

I built Ideanation as a Rovo app tightly integrated with Confluence and Jira. Rovo reads page and ticket context, uses structured prompts to generate consistent artifacts, and writes results back into the system of record.

The workflow is intentionally human-in-the-loop: users approve and refine outputs before moving forward. Each step builds on the previous one, maintaining traceability from problem → persona → story → requirements → BDD scenarios.

The focus was not on building a new UI, but on embedding intelligence directly into existing Atlassian workflows.

Challenges I ran into

Trying to get everything to work within Rovo without resorting to forge-llm
When moving a large amount of tickets over the agent would stall (though just checking in with it gets it to continue)

Accomplishments that I'm proud of

Built an end-to-end workflow from problem discovery to Jira-ready specs
Building the boat while sailing on it (the Jira spec requirements aspect was created first and then used to build the rest of the application)

What I learned

Platform specific AI implementations adds tremendous value to workflows by making it easier to consistently use AI

What's next for Ideanation

More customization (allow users to specify what frameworks to use to generate requirements etc)
Would like to use this on mobile (the actions don't seem supported on confluence mobile at the moment)
A whole ton of use. Win lose or draw this project is already helping me build a lot of other apps.