How incident.io relies on Recall.ai for their AI SRE

Executive Summary

Challenge

incident.io is an end-to-end incident management platform that helps teams respond when something goes wrong, coordinate responders, communicate updates, and learn from incidents afterward.

Conversation data is critical to that workflow. incident.io already had visibility into what happened in Slack messages, but when responders moved onto a call, much of that context could disappear unless someone manually took notes.

The team had wanted to automate call notes for years, but supporting the different meeting platforms its customers used would have required building and maintaining separate integrations.

Solution

incident.io built Scribe, its real-time incident note-taker, on Recall.ai.

Recall.ai gives incident.io access to conversation data from meeting platforms through one integration, allowing Scribe to transcribe incident calls in real time and bring that data into the broader incident record.

That lets incident.io focus its engineering resources on using incident data rather than building and maintaining meeting infrastructure.

Results

By building Scribe on Recall.ai, incident.io:

• Gave customers a more complete incident record by combining context from Slack messages with conversations happening on calls
• Replaced manual note-taking during incidents with real-time transcription
• Freed engineers to focus on resolving incidents rather than documenting them
• Saw roughly 60% of customers using Google Meet or Zoom enable Scribe within months of launch
• Transcribed tens of thousands of hours of incident calls within the first few months
• Made Scribe one of incident.io’s fastest-adopted features
• Used the same conversation data to accelerate broader AI and investigative-agent functionality

Capturing the full incident context across Slack and calls

incident.io helps engineering teams manage the full lifecycle of an incident, from coordinating the response through communicating updates and learning from what went wrong.

To do that well, context matters.

Some of an incident happens in Slack messages. Much of it can also happen once responders jump onto a call, where engineers are forming hypotheses, sharing findings, making decisions, and coordinating next steps.

Before Recall.ai, incident.io could see the messages in Slack, but the call context could be lost unless someone documented it manually.

“When it comes to building an amazing incident management platform, context is everything,” said Pete Hamilton, co-founder and CTO of incident.io. “In a world where half the chat is in Slack [messages] and half the chat's on a call, before Recall, half the context was missing.”

Recall.ai gave incident.io a way to capture that missing layer.

“Now whether the conversation's happening in Slack [messages] or on a call, we see the whole thing,” Hamilton said.

Automating incident notes without building every meeting integration

incident.io had wanted to automatically capture notes from incident calls for years.

The challenge was supporting the different meeting platforms its customers used. Building directly would have meant creating and maintaining separate integrations for providers such as Google Meet and Zoom, then repeating that work as platform coverage expanded.

“We could build an integration with Google, we could do one with Zoom, but there's loads of these things and they're all different,” Hamilton said.

That was not where incident.io wanted to spend its engineering resources.

The company specializes in taking large amounts of unstructured incident data and turning it into information that helps customers resolve incidents faster.

“Our specialty is not is building deep and rich integrations with an endless list of call suppliers,” Hamilton said. “I want my team figuring out how to use that information, not building those integrations.”

Recall.ai gave incident.io a single integration for the meeting recording layer, turning an idea that had previously required too much engineering work into something the team could realistically ship.

“It’s less like before I had two teams working on this and now I get it for free,” Hamilton said. “It’s more like you [Recall.ai] made a product possible.”

Turning incident calls into usable data in real time

incident.io used Recall.ai to build Scribe, its automated real-time note-taker for incident calls.

Scribe captures what responders are discussing without requiring someone on the engineering team to manually document the conversation. It can also help keep stakeholders informed without requiring everyone to join the incident call.

Scribe allows incident.io to “in real time forward updates to key stakeholders and let them know what's going on without having to have unnecessary people in the chat,” Hamilton said. As engineers hop off-call and new engineers replace them, the incoming team is immediately prepared to handle an incident without repeating the work that the previous engineers did.

Afterward, the same record gives teams a source of context for understanding what happened during the incident rather than relying entirely on notes written in the middle of the response.

For incident.io, that turns the call from an isolated conversation into usable incident data.

Keeping senior engineers focused on resolving incidents

Manual incident notes have a direct engineering cost.

During a major incident, the people on the call are often some of a company’s most experienced engineers. Having one of them primarily document what others are saying takes capacity away from diagnosing and resolving the issue.

Historically, Hamilton said, teams might have several senior engineers on a call with one “basically just there to take notes.”

“With the introduction of Recall, we've been able to essentially replace an entire person's worth of capacity in every incident by taking the notes for you,” he said.

incident.io transcribed tens of thousands of hours of incident calls in just the first few months through Recall.ai.

“That’s tens of thousands of hours of their senior engineers' time, which is incredibly precious,” Hamilton said. “This is not just a convenience, it's a meaningful business impact.”

Scribe, powered by Recall.ai, became one of incident.io’s fastest-adopted features

Customers quickly validated the need.

Within months of launch, roughly 60% of incident.io customers using Google Meet or Zoom had enabled Scribe.

Hamilton described it as one of incident.io’s fastest-adopted features to date.

The qualitative response was equally strong. Customers told incident.io that Scribe was among the best features the company had shipped that year, with some telling the team, “you'll never take this away from my team now.”

The adoption reinforced that automated incident notes were not simply an infrastructure improvement. They addressed a problem customers already felt during every incident.

Conversation data is becoming a foundation for incident intelligence

Scribe is also part of a broader data strategy.

incident.io already uses large amounts of unstructured information to help customers understand and respond to incidents. Adding call data gives the platform a more complete view of what happened, even when the response moves between Slack messages and a live call.

That has helped incident.io accelerate its broader AI efforts and bring investigative-agent functionality to customers sooner.

The shift is important: instead of treating the incident call as a standalone meeting, incident.io can treat what was said on that call as part of the incident record.

That gives the company more context to summarize what happened, communicate updates, support responders, and help teams learn from incidents afterward.

Why startups like incident.io shouldn’t build meeting integrations

For incident.io, the build-versus-buy decision ultimately came down to focus and speed.

Hamilton said building directly may make sense for a company that only needs one deeply integrated meeting platform and has the time to invest in it.

But that calculus changes when a fast-moving company needs to support several platforms at once.

“If you are a fast growing, fast moving startup and you have four platforms, five platforms to integrate with, you've got a million other things on your list and it all needs to be done tomorrow,” Hamilton said. “You just can't wait.”

incident.io also evaluated Recall.ai as critical infrastructure. The team looked at reliability and security because of the sensitivity of the data it handles, as well as responsiveness, speed to production, API quality, documentation, and client libraries.

Recall.ai allowed incident.io to ship Scribe without making meeting-platform infrastructure another core competency.

That leaves incident.io’s engineers focused on the higher-value problem: turning incident data into faster response, better communication, and a clearer understanding of what happened.