Hey Community! đź‘‹
I've been building Veronica, an AI-powered sprint assistant that lives right inside Jira. No external dashboards, no context switching — just an assistant that helps you Summarize work, draft replies, judge sprint health, and automatically chase down stalled issues. Here's a quick demo:
Demo video: Veronica. An AI powered Jira assistant
What Veronica does
Veronica shows up as a panel on any Jira issue, plus runs an autonomous bot in the background:
- AI Issue Summary — one click turns a messy issue into a clean structure: Summary, Progress, Risks & Blockers, and Next Action.
- Comment Reply Assistant — reads a comment thread and drafts three context-aware reply suggestions. You stay in control and just pick one.
- Sprint Data View — pulls the sprint an issue belongs to, its goal, state, and all its issues.
- Sprint Health Analysis — an AI verdict on the sprint (Healthy / At Risk / Critical) with at-risk issues, blockers, and a recommended focus.
- Auto-Escalation Bot — runs on a schedule, finds issues stuck "In Progress" for too long with no updates, posts an escalation comment, and opens a JSM ticket. It uses storage to avoid escalating the same issue twice.
The tech behind it
- Atlassian Forge — hosted and secured by Atlassian, no infrastructure to manage
- UI Kit for the frontend (native Forge components)
- Scheduled triggers for the autonomous escalation bot
- Forge Key-Value storage for escalation state / duplicate prevention
- JSM integration to create escalation tickets
- NVIDIA NIM as the LLM backend, called through Forge's declared egress controls
This is what I have created so, far in my preparation of Atlassian internship.
For next I have decided to make Aura is an AI-powered Confluence macro that brings live Jira sprint health into any Confluence page.
And then Pixie is an AI-powered Jira Service Management assistant that helps agents resolve tickets faster by surfacing relevant Confluence Knowledge Base articles right inside the issue panel.
