Loom, the asynchronous video app acquired by Atlassian last year, is extending its reach into synchronous video with the acquisition of Rewatch, announced this week. This brings an AI-powered assistant that is able to join virtual meetings in Zoom, Microsoft Teams and Google Meet, and will automatically record the meeting, take detailed notes, extract action items, and send follow-ups to meeting attendees. Once Rewatch is fully integrated into the Loom platform, these recordings and notes will be searchable across Atlassian products such as Jira and Confluence, and in future will be accessible to Rovo, Atlassian's intelligent agent technology, to automatically create updates to related Confluence pages, Jira issues, service tickets, and so on.
Until now, a big part of Loom's mission has been helping people to avoid spending time in video meetings, on the assumption that viewing a presentation that's been pre-recorded in Loom is often more convenient and efficient than watching it live. But live meetings still play an important role in moving work forward, so it makes sense for Loom to play a role there, too. Joe Thomas, Co-Founder and Head of Product Management at Loom, sees the two modes as complementary:
Asynchronous and synchronous work are like yin and yang. So it's an 'and' statement to us. We know that, even if you are the optimal version of utilizing async video, like Atlassian and Loom is, there is still a healthy amount of number of hours per week that you spend in meetings. So how do you think about getting the most value from those meetings? [How do you] structure the meetings in such a way that, during that time, it's not about information dissemination, it is all about collective creativity?
The ideal, he goes on, is to record video content on Loom so that attendees can get up-to-speed on the context before the meeting starts, then use the live meeting time for discussion and collaboration. Having an AI assistant there to record the meeting, make notes, and create follow-up actions, leaves attendees free to focus on the discussion. He says:
What do you do with that meeting content? A lot of important discussion topics were had, and a lot of the onus is on the note takers. Did they take the notes accurately? Is it comprehensive?
We believe that there's a lot of value in pairing up AI with meeting recordings, such that you can have a note taker there on your behalf [that] calls out key action items.
The promised integration with Rovo intelligent agents is the icing on the cake that differentiates the Loom assistant from AI assistants that are already provided by the meeting platforms themselves. Those native AI assistants don't have the outbound connections that allow them to automatically create or update content and tasks in Confluence or Jira directly from the meeting notes. He explains:
The [native] AI that wraps around it is accurate and it's valuable, but it sits in a silo. The way that work actually flows across the organization... a lot of work gets done in Confluence, and it gets done in Slack, and it gets done in Jira. Right now, that [native] content, the AI summaries, you have to go to the web page in order to read that and consume that, and it doesn't tie back into the rest of your work. With the Atlassian platform, with Rovo, we can actually turn that content into more value and agentic AI, in order to get things done.
It's this ability to automatically create or update content and tasks in Confluence or Jira directly from the meeting notes that customers are expected to find most useful. He gives an example:
The Jira team and Bitbucket team are working together to say, can we go from a Jira issue to an actual PR [pull request] that's opened on your behalf? Again, humans will manually check that that code is accurate and that it will solve the problem that it's looking to solve. But with the stack that Atlassian has, we have opportunities to solve unique things that other [vendors] may not be able to solve.
You could imagine, in a meeting recording, that we're talking about a specific front-end issue -- there's a bug that exists. Can that automatically create a Jira issue? Then, does that tie into the rest of Rovo agents to say, 'Oh, let me actually just open a PR for you and solve this issue.' So right when you're done with the meeting recording, we have a PR open and it resolves it.
I just think that there's unique, deep workflows that Atlassian can solve relative to the other platforms, and that is why we're continuing to push deeper into this space.
While there's an obvious benefit when the meeting is about work that's being progressed elsewhere within the Atlassian platform, he believes that Atlassian customers will often find it useful to have the Loom AI assistant listen in on meetings about other issues such as HR or sales. He explains:
If you're already deeply integrated into the Atlassian ecosystem, Rovo also creates a unified search. You could imagine certain pieces of a meeting or an entire meeting itself, is valuable for employee onboarding. Why wouldn't you want to make that content available in Atlassian search? That way it shows up whenever, even if you're in Confluence or Jira, and you go to our unified search offering.
The notes created by the AI assistant use templates for common meeting types, and users can edit them or add their own notes. Transcripts are searchable and can be copied in parts or downloaded in full, with translations available in over 50 languages. Integrations with both Google and Outlook calendar apps allow users to schedule the assistant to join and record meetings, even if the user doesn't attend in person.
Actions created by a Rovo agent will always be reviewed by a human before going ahead. Thomas says:
We're still many years out from having that be fully automated. There's a human checkpoint at each point of the workflow that says, 'Hey, is this Jira issue accurate?' ... I do think that there will be human checkpoints and 'interventions' at each stage for the foreseeable future.
Atlassian's perspective is that it's the role of AI to help people work faster and smarter, rather than replacing them. He concludes:
AI ... is for augmentation. It is up-leveling our ability to do more strategic, more creative, deeper work. Our philosophy is not to get rid of the human interventions or checks for accuracy. It's hopefully to help them up-level their work generally.
Although Loom has had the ability to import Zoom recordings for some time, Rewatch brings a much richer set of capabilities across all the major virtual meeting platforms, and once the Rovo integration has been delivered, you can envisage that this will offer Atlassian customers a lot of value. However to realize that value, an enterprise must already be a big user of Jira and/or Confluence.
Most organizations therefore will probably use the Loom assistant in addition to, rather than in place of, the native AI assistants in those virtual meeting platforms. Indeed, when you think about it, there's no real reason not to have several different assistants listening in to each meeting, ready to pick up and complete the actions that each of them are configured for -- Loom taking actions back to Atlassian for development workflows, others taking actions back to their own work management platforms for workflows in HR, finance, sales and marketing, and so on.
Extend that further, and you could also make a case for putting a microphone into a wholly in-person meeting, just so that the AI agents can take care of the meeting notes and follow-up actions.
Look at this announcement, therefore, as a sign of the increasing digitization of work on the path to Frictionless Enterprise, in which adding a digital recording to a live meeting allows the proceedings to flow directly into workflow automation, stripping out much of the old manual friction of note-taking, minutes and follow-ups.