Stay in the meeting. Hey Athena keeps your work moving.

Hey Athena is a feature of Athena Intelligence that turns spoken requests in a meeting into follow-up work. Anyone in the room says "Hey Athena" and asks for something. Athena identifies who spoke, sends the request to that person's own account, and runs it only after they approve it. It is an AI agent for meeting follow-ups that keeps the meeting moving.

Meetings produce requests faster than anyone can act on them. This dispatch covers what Hey Athena is, one request followed from the room to an approved result, and the controls that keep every request tied to the person who made it. Figure 1 shows the whole flow on one page.

01What problem does Hey Athena solve?

The follow-ups from a meeting land on whoever types fastest, while the discussion moves on without them. In a survey of 5,000 knowledge workers, Atlassian found that 54% frequently leave meetings without a clear idea of next steps or who owns which task.

Picture a quarterly business review with forty people in it. Someone asks for churn by region for the board pre-read. Someone else promises a customer follow-up. A third person asks who owns the pricing memo. Each ask goes into chat, a notes doc or an inbox, and one person ends up chasing all of it across the notes, the inbox, a model in Excel and the pre-read itself. Microsoft's Work Trend Index found that 55% of people say the next steps at the end of a meeting are unclear.

02What can you ask Athena to do in a meeting?

Anything you would otherwise type into chat or add to a follow-up list. Athena drafts the work, and the person who asked reviews it. Figure 3 groups common requests by team.

For leadership and board prep, that means pulling a number for the pre-read, assigning an owner to an open memo, or answering a factual question from your own documents without anyone leaving the call. Sales, RevOps and customer success teams can have the customer follow-up drafted, the account record updated, action items filed against owners and the next meeting proposed. See the Sales, RevOps & CS solution page for more on that work.

Finance teams can ask for churn, revenue or variance by segment and get a draft that notes which model the number came from, ready to check before it lands in the deck. Operations and project teams can capture decisions and owners as they are spoken, then send a recap for approval instead of rebuilding it from memory. Anyone can ask what was said about a topic across past calls and get the answer with a timestamp, and a new hire can catch up on an account's call history in an afternoon.

03How does it work? One request, start to finish

Someone says "Hey Athena," and the request reaches the right person's screen, ready to approve. The video follows one request through four steps, and Figure 4 lays them out.

1. During the review, Daniel says, "Hey Athena, pull churn by region for the board pre-read." The meeting keeps going, and nobody opens a laptop.
2. Live speaker identification matches each voice to a teammate, so the request belongs to Daniel and not to the person recording.
3. After the meeting, the request appears on Daniel's screen in his own account, with his words transcribed from the recording.
4. Daniel can submit, edit or decline. Nothing runs until he submits, and unanswered requests expire at the end of the day.

When Daniel opens the session, it starts with his exact words and a link to that moment in the recording, so anyone can check what was actually asked. Table 1 puts the manual version and the Hey Athena version side by side.

04Who stays in control of each request?

The person who asked. Their request runs in their own account, under their own permissions, and only after they submit it. Figure 5 shows the request card, and Figure 6 shows where each step happens.

Because every request carries the words that were spoken and a link back to the recording, checking it takes a few seconds. The person can submit the request, edit it first, or decline it. Requests nobody answers expire at the end of the day, which keeps old asks from piling up or running later without anyone noticing.

Whoever starts the recording holds no authority over what other people ask for. Each request belongs to the person whose voice it was. After they submit, Athena opens a normal chat session in the asker's account, starting with their exact words and a link back to that moment in the recording. From there the person can keep working in the chat, or run an agent operating procedure (AOP) from it. Athena runs inside your environment, next to your data (see deployment options), and more on how every step stays attributable is in attribution and reversibility.

05How is this different from other AI meeting tools?

Most AI meeting tools stop at the summary. Athena treats the transcript as the place where work starts. Meetings become a source of drafted follow-ups, and the notes are one of the things Athena produces along the way. See Athena for Meetings for the full picture.

The person who asked stays in control, because requests run in the requester's account under their permissions and only after they approve. Every request can be traced to the exact words and the moment in the recording. And everything sits on one platform: meetings, documents, data and workflows live in the same place, so a request like "pull churn by region" can reach the right model or dataset without another tool. The voice and mobile surfaces extend the same idea beyond the meeting room.

06Is Hey Athena secure enough for regulated companies?

Yes. Athena Intelligence is SOC 2 Type II and HIPAA compliant, has zero-data-retention agreements with Anthropic, OpenAI and Gemini, and does not train on customer data. Athena deploys in your VPC, on-prem or air-gapped, works with 60+ models, and has been in production at a Big Four audit firm for over a year, with over 1,000 users.

07How do teams get started?

A good first meeting is one that recurs and where follow-ups tend to slip, such as a weekly business review or a customer call. Hey Athena is part of Athena for Meetings, so the recordings, transcripts and speaker matching are already in place.

From there, the same approach reaches finance, sales and operations teams, since every request runs on the same platform with the same approvals. Curious how it would work in your meetings? We're happy to show you.

08Questions teams ask