Budget tracking
A live backlog, matched against the original estimates.
A member of the Mirego team built Split Up in Forra to pull Jira issues in real time and match them against the estimates the project started with.
The challenge
The gap widens while the project moves forward
In a development project, the initial estimates live in a document, but the actual backlog is in constant motion: tasks get refined, split and added week after week. Knowing whether you are still within the budgets requires tracking that nobody really has time to do manually.
A team member had cobbled together a solution in a Google Sheet with formulas, but keeping it in sync with what was actually in Jira was not easy. Every reconciliation meant exporting Jira, lining tickets up against the estimate document, and rebuilding the picture from scratch. That is usually where the tracking quietly stops.
The solution
Split Up: an application that redoes the reconciliation on its own
A team member built Split Up in Forra to make that link automatically. Thanks to the Jira skill, the application pulls tasks in real time and matches them against the estimates in the original document, using AI to establish the connections between two lists that never shared an identifier: on one side tickets that have been split and renamed along the way, on the other estimate lines written before the work began.
The team member iterated for about two hours to refine the result, with targeted adjustments each time. The Jira skill structures the call to the external tool rather than letting the model interpret the API on its own. Data comes back structured the same way every time, and scoped to what the person opening the application is already allowed to see in Jira. No integration code was written for any of it.
The outcomes
What the team gets out of it
- The project manager and the product owner get a clear, up-to-date view of the project against what was planned.
- Conversations with the client rest on the real state of the project, helping the team make informed decisions as work moves forward.
- No Google Sheet to keep alive alongside Jira.
Eddyfi’s knowledge base already had the answers. Finding them was the hard part.
Purpose-built AI agents replace a generic assistant, achieving 87.5% accuracy on a validated benchmark versus 12% for Microsoft Copilot.
The power of experience: capturing 30 years of knowledge.
AI transcribes expert interviews, identifies speakers, and turns institutional knowledge into a searchable resource for your entire team.
The answer already existed; reaching it just took 15 hours.
AI research agent searches four scientific libraries simultaneously and delivers a complete, sourced report in minutes instead of hours.