Loom Record for agents - Hit record and hand off to AI, now in the new Loom app

Hi Product Hunters!

I’m Briana Ings from the Loom team at Atlassian, and we couldn’t be more excited to introduce Record for agents, built directly into our brand new Loom for Mac desktop app!

The Problem: Prompting AI is time consuming & misses the mark
AI agents are incredible, but prompting them with text alone can be time consuming. Typing out complex multi-step workflows or comparing Figma specs for a design review can take forever, and key context can easily get lost. When agents miss those visual details, they often build the wrong thing, leaving you stuck re-prompting and redoing the work. Because of this, we found ourselves spending more time typing walls of instructions and taking screenshots than actually getting work done.

Show, Don’t Type: How Record for agents works

Instead of writing endless text prompts, you can simply record your screen and talk through what you want done.

Loom captures your voice, actions, URLs, and key visual moments (using native desktop tools like the Magic pen and Selection tool), then transforms the walkthrough into a clean, structured prompt ready for AI agents like Rovo, Claude, Cursor, or ChatGPT to run with.

To power this, we completely rebuilt Loom for Mac for macOS 15+. It’s faster, has an 80% smaller footprint, and brings new camera styles, area-specific recording, teleprompter style Speaker Notes, and so much more.

Record for agents turns spoken intent and on-screen context (clicks, element selections, keyframes, console/network logs, and links) into structured, multimodal instructions for AI agents (Rovo, Claude, Cursor, JCA, etc.).

Where Record for agents shines:

When a task is tedious or slow to explain in text, just hit record:

• Slide decks & presentations: Talk through an all-hands or monthly review while pointing to live dashboards and source docs—let your agent draft the slide copy and pull the visuals.

• UI tweaks & bug fixes: Point directly to visual bugs, layout quirks, or styling changes so your coding agent knows exactly what to fix without guesswork.

• Docs, briefs & summaries: Flip through multiple tabs, project boards, and data sources to generate a structured brief or status report in seconds.

• Prototypes & user flows: Walk through a desired user journey across apps to give agents full end-to-end context.

• Feature specs & PRDs: Brainstorm and narrate your idea out loud while referencing existing tools to produce a comprehensive spec.

Ok this one I'll actually use. Half my agent runs end with "wait, what did it even do" and a Loom of the session fixes that. Nice one.

We're really excited to release this and the all new native Loom for Mac. Recording a Loom and simply showing your agent what you want is incredibly powerful, and we're just getting started. Let us know what we can add or do to make it fit better into your workflow.

Hi Product Hunters!

Omri from the Loom team and excited to share the new Loom mac app with you!
We're here to make Loom the best we can with your helpful feedback!

We build Loom on Loom in a few concrete ways, for things you can try with your team:

1. Everyone doing 'show and tell' by recording Loom demos and sharing on slack to spark discussion and get alignment

2. "This meeting could have been a Loom" - avoid unnecessary meetings and get maker time back by sharing our thoughts with each other asynchronously over Loom. This is more inclusive of time zone, team members different processing style and leaves the time when we do meet to be of high value.

3. Building with AI like we collaborate with a teammate. Instead of 'mastering the art of the prompt', we just record our thoughts and point our mouse and screen at what we want to build, and the models take it from there. This is also helpful for creating documents, slides or other content from your thoughts.

4. PR walkthroughs in Loom - When we open PRs, the authors often record a Loom with their thoughts on the change, context and considerations. This makes PR reviews a lot more human and effective, and helps build our 'institutional memory'. Since all of our Looms are recorded in Atlassian's team work graph, its very easy to find context that might be buried in that PR 8 months ago.

5. Collaborate - Comments on Loom are accessible/editable by the MCP. So if there's any feedback on a Loom, you can pull it into any of your workflows to address.

Can’t wait to hear everyone’s early feedback. Please leave a comment so we can share positive thoughts, feedback, and ideas with the team!

Congratulations on getting this out team!