# Dreamwalk

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Prasanna Rajendra Vaddkkepurakkal](https://cerebralvalley.ai/u/prasannavadk)
- **GitHub:** https://github.com/prasannarajezzzy/engineering-copilot-widget
- **Demo video:** https://youtu.be/dR4Az6p_bRk
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4055

DoLittle is a desktop engineering copilot that turns meetings and tickets into real work.

- **The problem:** engineers don't lack AI, they lack continuity. Work is split across five tabs, meeting notes die in docs, and tickets get retyped by hand.
- **The golden path:** one small widget listens to the meeting, produces a summary with decisions and action items, and takes an action item — or a live Jira ticket — into Task Flow, where an agent shapes it into a single Markdown task file written into the project folder and handed to Cursor or Claude, ready to build.
- **Context travels with the task:** Brief shows the live Jira dashboard, Tickets loads your Jira project, and Projects links each GitHub repo to its local folder, so every task knows where it belongs.
- **Built for daily use:** open it every morning for your daily standup, your project meetings, or any working session, and it turns each one into a summary with next steps — so action items stop sitting in docs waiting to be retyped.
- **The manual cost it replaces:** note-taking and chasing down action items can eat around 1–2 hours of a workday, and it is easy to forget to open an AI companion and ask it anything — so things simply get missed across a day of meetings. DoLittle records your daily tasks and action items as the meetings happen and keeps track of them, so nothing falls through.
- **Jira in the same flow:** it pulls your Jira tickets in and turns meeting notes into ready task files.
- **The prompting win:** normally your first prompt to a coding agent is blank, which is why so much iteration follows. DoLittle's built-in task draft gives the agent a context-ready Markdown file supported in Cursor and Claude instead. In my own daily use, I get around 70% success on the first chat attempt with that task file, versus starting from a blank prompt.
- **Proven in the demo:** the full path — listen, summary, Task Flow, task file, handoff to Cursor — runs end to end in under 3 minutes, in one widget instead of five tabs, with no retyping between tools.

## More from Meta Global AI Developer Hackathon

- [Kitsa](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4052)
- [CLEVERS](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4053)
- [Agentic Corporation](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4054)
- [WonderLab](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4056)
- [Tinker](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4057)
- [Lab2Rob](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4058)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4055. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt