←
Product Design
Motion Design
AI
Fletch
Created an in-context companion, surfacing related Looms instantly on a Confluence page
Role
Product Designer
Team
2 Product Designers
2 Engineers
Timeline
24 Hour Hackathon
The Background
Bringing Related Looms to Confluence Pages
Fletch was built during Atlassian's internal intern hackathon (ShipIt), where interns were given an open brief — tackle any problem across the Atlassian ecosystem. With the entire product surface available to us, we zeroed in on a gap hiding in plain sight: the disconnect between where teams record knowledge (Loom) and where they actually work (Confluence).
HMW make Loom videos discoverable to people who don't know they exist?
The Problem
Why Loom?
Less than 2% of Loom videos are searchable or discoverable. Users can generally only find Looms they created or were directly shared with — meaning the vast majority of recorded knowledge is invisible to the organization
The Solution
Introducing Fletch
Here is our demo video previewing how our solution, Fletch, surfaces relevant Loom videos based on the information of the Confluence page you are viewing.
User Interviews
Understanding what current users actually need
Given the compressed hackathon timeline, we leveraged existing Atlassian UXR as our research foundation. Internal studies across 9+ user interviews and multiple synthesis reports consistently identified video discoverability as the #1 viewer pain point
"Unless I have a link to a Loom I find it really hard to find them. I have to go back through my pings with [my boss], where did she send me that link?"
"I would assume that…if the video is open…in Loom [or] where people are used to search, I think you would assume that, okay, it's open, it's searchable."
Final Solution
In-Context Companion
Fletch surfaces relevant Loom videos directly inside Confluence — no searching, no tab switching, no dead-end share links. Powered by Teamwork Collection (TWC), it reads the context of your current page and instantly fetches the most relevant videos, each with an AI-generated overview, turning buried organizational knowledge into something your whole team can actually find and use.
Outcome and Learnings
Key Takeaways
Ground the problem in real research — validating the problem with data early on made it easier to communicate business value and get buy-in, helping us place 2nd overall.
Align with engineering early — with limited hackathon time, working closely with engineers on feasibility shaped what the solution could realistically look and function like.