With the new Loom for Mac, Record for agent, and MCP tools, teams can now hand off work to AI agents the same way they do to teammates: by showing the work, with every detail captured.
Explaining an idea clearly enough for someone else to act on used to be one of the slowest parts of collaboration. Loom video messages changed that for teams: instead of typing it all out, they could show it instead. What began as a quick screen recording grew into recorded meetings and narrated walkthroughs, until “just send a Loom” was shorthand for showing someone what you mean. A quick video could align remote teams across time zones, efficiently provide feedback, and foster human connection asynchronously. In a 2025 survey of 7,000 knowledge workers by Loom and Wakefield Research, 70% said a short recorded video message would have prevented their most recent miscommunication.
As more teams turn to AI agents to help build, debug, create, and ship customer value, they are running into the same communication bottleneck. Getting an agent the context it needs means typing out prompts, uploading screenshots, and copying links between tools. Any ambiguity in the prompt adds time and back-and-forth until the agent gets it right. And the more visual, nuanced, or complex the task, the more detail gets lost in translation.
Handing off work is already hard enough between people. And with AI agents in the mix, there’s even more room for error and more chances for quality to slip. In Atlassian’s The State of Teams 2026 report, 87% of knowledge workers say they lack the time or capacity to coordinate, and as one Fortune 500 CTO put it, “With agents, coordination will get worse before it gets better.”
Now working with agents is as easy as firing off a quick Loom to a teammate. With Record for agent, the new Loom for Mac, Loom tools in Atlassian MCP, and Rovo in Loom, teams can brief AI more clearly, put the context from every recording to work across their tools, and ultimately deliver more value to customers.
Think outside the text box
Handing work to an agent breaks down in the same three places it does with teammates: getting intent across, carrying context between tools, and understanding organizational knowledge. Loom now addresses each.
Record with Loom, hand off to any agent
Record for agent is a new way to record in Loom that turns short screen recordings into high-fidelity prompts for AI agents.
As you record, Loom captures the key context your agent needs: the transcript with your narrated instructions, screenshots of on-screen elements you clicked, and links you navigated to while sharing your screen. It then turns all of that into a clear set of instructions the agent can act on.
In the new Loom for Mac, a rebuilt desktop app with sharper, 4K-by-default recording, two tools make capturing intent unmistakable: the Selection tool pins a precise UI element for the agent, and the Magic pen adds freehand annotations on screen to call out a shape or area.
From there, hand the prompt off for any agent to execute by sending it directly from Loom, or opening the prompt as a work item in Jira that can be assigned to an agent.
See it in action
It works for any task that’s slow to put into words. Here are a few examples to start:
- Slide decks & presentations. Talk through a monthly update or all-hands deck, pointing to source docs or live apps to have AI draft your slide copy and grab screenshots.
- UI changes & bug fixes. Point directly to elements on screen so an agent knows what to adjust.
- Docs & summaries. Walk through several open tabs, dashboards, or project boards to generate a structured report or brief.
- Prototypes & user flows. Navigate across apps to explain a desired user journey or workflow.
- Feature specs. Talk through an idea to generate a comprehensive technical or product spec.
Unlock your team’s video intelligence from any AI tool
Loom now supports the Model Context Protocol, so recordings stay a source of context to the agents and tools your team uses, long after the handoff.
Loom tools in Atlassian MCP let agents like Claude, Cursor, VS Code, Codex and Rovo work with Loom directly. An agent can pull a video’s transcript, metadata and action items, add timestamped comments, and share or organize recordings, all within a team’s existing Loom permissions. Loom tools are built directly into Atlassian MCP, so there’s no separate server to configure.
If you already use Atlassian MCP v2, Loom tools are available automatically once your Loom workspace is linked to your Atlassian site. If you’re currently on v1, upgrade to Atlassian MCP v2 by updating or reinstalling the Atlassian connector in your AI client to unlock Loom capabilities.
Get started with Loom in Atlassian MCP →
Rovo in Loom, rolling out now, puts your team’s recordings to work without leaving Loom. Rovo Search finds the right video or moment across a library, and Rovo Chat answers questions about what was said, drafts the follow-up, or turns a walkthrough into written steps. Each answer is grounded in the recording and exact timestamp, so you can get the key highlights, without rewatching the full recording. Connect other tools and it searches those too, all within the team’s existing permissions.
From idea to shipped change
Take one example of how to use Loom to work with agents: shipping a UI change.
Instead of writing a spec, record a Loom and walk through the current behavior, what should happen instead, and edge cases, pointing at the exact elements on screen. Loom captures the screens, clicks and links, so the agent prompt paints a rich picture of the ask, without a single screenshot taken by hand.
From there, hand that prompt to a coding agent, or connect to Atlassian MCP and let the agent pull it straight from the link. Once connected, the agent is grounded in the permission-aware context powered by the Teamwork Graph. So when it gets stuck, have it can surface fixes the team already shared, instead of spinning on a problem solved last month. When ideas change, or new requirements come in, a ten-second “not that button, this one” Loom is all it takes to course-correct.
The same Loom that prompts the agent becomes a resource teammates rely on when they review the result. The team has full visibility into what the agent was instructed to do, judge the result against it, decide when it’s ready for customers.
The result? Less time explaining and more time building. Agents handle the first draft, which leaves people free to cus on the judgement calls that make the work high-quality and get it to customers faster.
Start using Loom with AI agents
For years, “just send a Loom” was the fastest and clearest way to show a teammate what you mean. Now it’s the best way to show an agent, too. Record for agent, the new Loom for Mac, Loom tools in Atlassian MCP, and Rovo in Loom all work toward the same thing: show your intent once, and it carries across the AI tools your team already uses to ship high-quality work.
Record for agent is in open beta for Loom Business + AI and Enterprise plans on both the new Loom for Mac or the Loom Chrome extension. To let agents work with Loom from any tool, connect Loom tools in Atlassian MCP.
