Honeycomb Surpasses Company Growth Milestones as Customers Turn to Observability to Understand and Manage Swarms of AI Agents
Company sees spikes in AI product adoption as engineering teams use Honeycomb to investigate and observe their AI agents in production.

By: Ciri Haugh

Company sees spikes in AI product adoption as engineering teams use Honeycomb to investigate and observe their AI agents in production
SAN FRANCISCO – November 18, 2026 – Honeycomb.io, the observability platform built for the AI era, today announced a record-breaking year of company growth, product innovations, and customer adoption. Over 70% of Honeycomb's customer base actively relies on Canvas or the Honeycomb Model Context Protocol (MCP)—key elements of the company's AI features—every week, with a 30x surge in MCP tool calls since February. Over the past four quarters, Honeycomb's ARR grew faster than in any prior four-quarter period, up more than 70% year over year. The company also formally entered the AI SRE market, and grew its AI-native company customer count by 130%.
“Complex systems are where Honeycomb thrives, but AI made production harder to understand and the questions harder to predict,” said Christine Yen, cofounder and CEO at Honeycomb. “Canvas and MCP were our first answers. About three in four customers have used each this year, and that told us where to go next: help teams observe the agents they run, and put agents to work on investigating them. Each step has to rest on high-cardinality data, because an agent working from shallow data will only serve to give you a wrong answer quickly.”
Investing in observability for the agent era
Honeycomb believes that organizations need to understand their systems quickly, solve problems with confidence, and scale without runaway costs. AI agents now require a new approach to understand and manage behaviors, performance, and reliability issues. Developers and engineers need observability tools built to handle this shift.
According to Gartner, global spending on AI infrastructure, services, and software will reach $2.7 trillion in 2026, with more than $400 billion going to AI software alone.* As investment flows into AI, organizations have been turning to Honeycomb to help them manage the growing complexity AI adds to their systems.
Existing customers are expanding faster than in previous quarters or years, as net revenue retention (NRR) rose to 125% and gross retention revenue (GRR) grew to 89%. Honeycomb has also significantly increased the number of Fortune 500 and Global 2000 customers using its platform. In Q3 alone, 40% of Honeycomb's new enterprise customers were AI-native companies. Notable brands that use Honeycomb include Baseten, Conductor, Deloitte, Duolingo, ecobee, Fin, HelloFresh, Nubank, Strava, Zopa Bank, and Zurich Insurance Group.
“Honeycomb has been at the cutting edge of observability across some of the most complex software domains,” said George Mathew, Managing Director at Insight Partners. “Their continuous investment in AI observability, built on a decade of expertise in this market, we believe, is industry-defining. The passion, experience, and innovation the team brings to the observability space are notable, and Insight is thrilled by Honeycomb's continued momentum.”
In order to extend this record growth, Honeycomb recently promoted Matt Nelson from Chief Revenue Officer to President. He is now responsible for the entire Honeycomb go-to-market organization, with a focus on rapidly accelerating the market's adoption of the company's new AI SRE offering and further expansion of its enterprise customer base.
Tackling complex problems with innovation
Over the past year, Honeycomb has released a series of observability features built for the new production problems that AI agents create. AI is non-deterministic, so traditional best practices for debugging, incident investigation, reliability work, and other standard engineering and DevOps tasks no longer hold up.
The work started with the Honeycomb MCP, released in late 2025, followed quickly by Canvas, a multiplayer workspace where teams investigate issues and understand how their production systems perform. Among paying customers, 74% are actively engaging with the features weekly as a part of Honeycomb Intelligence. In fact, MCP-driven queries against Honeycomb grew more than 200x from January to October. AI agents and the MCP now run over 4.5 million queries a month against Honeycomb, up from about 20,000 in January.
Other significant product innovations include:
- AI site reliability engineering (SRE) agents: The newest release, Team Agents, works proactively towards goals set by engineering teams to monitor deploys, improve instrumentation, investigate alerts, improve agent quality, and more.
- Anomaly Detection: Understand and set baselines for normal activities, with the ability to set alerts or flags when error rates or presence go below an acceptable level.
- AI Ecosystem: Understand and manage a fleet of agents, with the ability to go deeper into performance, cost or behaviors.
- Honeycomb Private Cloud: Manage your observability in a dedicated private cloud, with Honeycomb-managed or self-managed options.
Honeycomb's technology partnerships extend these capabilities even further. This year, Honeycomb earned two Amazon Web Services (AWS) competencies: Financial Services and Resilience Technology, and took part in the public preview of OpenTelemetry tracing for the Claude API and Claude SDK from Anthropic. The organization also announced new or expanded partnerships with AWS, Anthropic, Cloudelligent, Checkly, incident.io, Improving, and Focused.
“Agent observability is evolving quickly, but understanding agents requires visibility beyond the model call,” said Austin Vance, CEO of Focused. “Honeycomb has spent years on the harder problem of following what software did and why, which is the right foundation for agents. That's why we build with them. When our clients put agents into production, we want a partner who already knows where the failures hide.”
To follow Honeycomb or learn more about its award-winning observability platform, visit www.honeycomb.io. Or find us at AWS re:Invent in Las Vegas, Nov 30 to Dec 4 to see any of these innovations or chat with us live.
*Source: Gartner, Forecast: AI Spending, Worldwide, 2025-2030, 2Q26, September 2026
About Honeycomb
Honeycomb is the observability platform for the AI era. Built on a decade of distributed tracing leadership and deep roots in the OpenTelemetry community, Honeycomb gives engineering teams—and the AI agents now operating alongside them—real-time, high-cardinality answers about any production system, with no pre-aggregation and no cardinality limits. Learn more at www.honeycomb.io and follow us on LinkedIn.
Media Contact
Ciri Haugh
Press@honeycomb.io


