Training Example: Honeycomb – Review the Data, Give Your Score & Compare to the Real AI Evaluation

[H1] Observability built for the AI era
Engineering teams use Honeycomb to follow their code into production. From tracing distributed services to debugging non-deterministic AI workflows, Honeycomb gives humans and agents a shared view of the only thing that matters: what's happening for their end users.Get a demoStart for freeTrusted by innovators and enterprises worldwide
[IMG: Slack]
[IMG: Intercom]
[IMG: Dropbox]
[IMG: Booking.com]
[IMG: Heroku]
[IMG: duolingo]
[IMG: Hello Fresh]
[IMG: jack henry]
[IMG: Slack]
[IMG: Intercom]
[IMG: Dropbox]
[IMG: Booking.com]
[IMG: Heroku]
[IMG: duolingo]
[IMG: Hello Fresh]
[IMG: jack henry]
[H2] Built for answers, not tradeoffs
From tracing distributed systems to debugging LLM behavior, Honeycomb helps you tackle today's most complex engineering challenges. It starts with the only purpose-built columnar data store for observability. You get the unified telemetry and AI agent integrations to go from alert to answer, before issues impact your customers.Explore our Platform
[H3] Send us all your data
Logs, metrics, traces–send any structured data our way without paying more for the additional data you capture. Add as much context as you want, and derive unlimited custom metrics for free. Our OpenTelemetry-compatible platform rewards curiosity without incurring extra costs.
[H3] Make your data your own
Define a strategy for your telemetry data that controls cost and helps you get the right answers now. Collect, enrich, filter, sample, route, and shape your data for faster troubleshooting and deeper insights.
[H3] A data store engineered for scale
Today’s software systems produce a volume and complexity of telemetry unlike ever before, and we custom-built our data store to handle it all. Append all the technical and business context you need, for no extra charge.
[H3] AI-powered observability with Honeycomb Intelligence
Make every engineer an expert and every investigation instant. Speed up your investigations with Canvas, an AI-assisted copilot, or access your observability data directly with your AI agent IDE via Honeycomb MCP so you can stay in the flow.
[H3] Answer questions in seconds, not hours
Say goodbye to slow queries and dashboards that lead to dead-ends. Whether you know exactly what you're looking for or you're just poking through the data, results from Honeycomb's query engine are unbelievably fast—especially when paired with Canvas AI copilot. Visualizations are dynamic and explorable, encouraging and empowering investigation.
[H3] Integrates with the tools you use
Honeycomb integrates with 60+ tools across the software development lifecycle, fitting seamlessly into your existing tech stack to keep you in flow. From CI/CD pipelines to incident management tools to AI-powered investigations, Honeycomb ties observability directly to the systems you already use.
[H2] What Honeycomb helps you do
The future of software is nondeterministic. Your business can’t be. Understanding production has never been more critical.
[IMG: See the whole picture, fast]
[H3] See everything
[H3] Get complete visibility with data available in under 90 seconds. Explore dynamic visualizations with limitless insights and no dead-ends. With Honeycomb, engineers and AI agents always have the context they need to easily solve problems.
[IMG: Move Fast with an MCP]
[H3] Solve anything
[H3] Resolve any issue before they impact customers. Run sub-10 second queries with your AI agent. Perform root cause analysis in under three minutes with BubbleUp. Honeycomb makes every engineer an expert and every investigation instant.
[IMG: Scale fast with first class OpenTelemetry support]
[H3] Scale infinitely
[H3] Stop making tradeoffs between observability cost and software quality. With Honeycomb’s OpenTelemetry-native platform, you can add unlimited fields and unlimited users at no extra cost, with no vendor lock-in. Honeycomb is ready for whatever comes next in your tech stack.
Thanks to Honeycomb, SLO-based monitoring has proven superior for LLM reliability than traditional metrics. Given AI’s unpredictable nature, SLOs help us catch and investigate anomalies without triggering a flood of noisy alerts.Kesha MykhailovStaff Engineer, Intercom, on Fin AIView Case Study
[IMG: Kesha Mykhailov]
[H2] Solutions for every team
We created Honeycomb to solve the problems legacy monitoring couldn't address. Explore the ways Honeycomb can work for you.
[H3] LLM Observability
Get granular insight into how your LLMs behave in production, troubleshoot failures faster, and continuously improve model performance—all in real-time with real data.
[IMG: LLM Observability]
[H3] Observability with AI Agents
[H3] Distributed Tracing
[H3] Metrics
[H3] Log Management and Analytics
[H3] Telemetry Pipelines
[H3] Private Cloud
[IMG: LLM Observability]
[IMG: Observability with AI Agents]
[IMG: Distributed Tracing]
[IMG: Metrics]
[IMG: Log Management & Analytics]
[IMG: Telemetry Pipelines]
[IMG: Honeycomb Private Cloud]
O'Reilly Book
[H2] Go deep on observability with the experts
AI is changing the nature of software development. Observability is how you navigate this shift. Charity Majors, Liz Fong-Jones, and George Miranda explain the core principles of observability and how to leverage it for your organization.Download your free copy of the bookGet a preview of what you’ll learnSign up for updates on the 2nd editionGet your copy
[IMG: Observability Engineering Book]
[H2] Keep learning...
[IMG: Everything We Talked About at O11yCon 2026]
Ken RimpleMay 27, 2026Everything We Talked About at O11yCon 2026Conferences & MeetupsSoftware Engineering
[IMG: Cloudelligent and Honeycomb.io announce a strategic partnership to help engineering teams modernize on AWS with deep production observability.]
Julie NeumannMay 27, 2026Cloudelligent and Honeycomb Partner to Bring Enterprise Observability to AWS-Native Organizations in the AI EraPress Releases
[IMG: SEO-BLOG_-What-Are-Traces_1]
Rox WilliamsJuly 11, 2025What Are Traces? A Developer’s Guide to Distributed TracingTracing
[IMG: Honeycomb Canvas: The Multiplayer Workspace for the Agentic Era]
Kale BogdanovsMay 20, 2026Honeycomb Canvas: The Multiplayer Workspace for the Agentic EraInnovation WeekProduct UpdatesAI & LLMsView all
[H2] Ready to get started?
Start for FreeGet a demo

[H1] Empower your team to work smarter and faster
See the whole picture Honeycomb’s data store unifies your data with the context you need to easily solve problemsMove fast and unbreak things Resolve issues before they impact customers with fast feedback loops and custom workflows.Observability for the whole team Every query and investigation is shareable and accessible forever in Honeycomb for everyone to explore.
[H6] Book your personalized demo
Trusted by innovators and enterprises worldwide
[IMG: Slack]
[IMG: Intercom]
[IMG: Dropbox]
[IMG: Booking.com]
[IMG: Heroku]
[IMG: duolingo]
[IMG: Hello Fresh]
[IMG: jack henry]
[IMG: Slack]
[IMG: Intercom]
[IMG: Dropbox]
[IMG: Booking.com]
[IMG: Heroku]
[IMG: duolingo]
[IMG: Hello Fresh]
[IMG: jack henry]
[H2] Loved by teams across the world
[H2]
[IMG: Pawel Malon]
After years in engineering feeling like I was missing critical insights, with our switch to Honeycomb, suddenly, it’s like I have sight for the first time in my career.Pawel MalonPrincipal Software Engineer, Phorest
[IMG: Edith Harbaugh]
We rely on Honeycomb to provide the world-class scale and resilience LaunchDarkly customers expect.Edith HarbaughCo-founder and CEO, LaunchDarkly
[IMG: Rich Anakor]
We’re able to bring a different mentality in the way we run and manage our production systems. We were able to really help our engineering teams and change the culture.Rich AnakorChief Solutions Architect, Vanguard
[IMG: Nick Herring]
People ask, ‘Where’s Honeycomb in the new architecture?’ The answer is it’s everywhere. It’s built in.Nick HerringTechnical Director of Infrastructure,
CCP Games
[IMG: Michael Garski]
We have predictable pricing that lets us do away with the legacy approach of sampling logs to preserve budget. Honeycomb’s event-based pricing empowers us to see every request from every service that comes through.Michael GarskiDirector of Platform Engineering, Fender
[H2] Ready to get started?
Start for free

[H1] Why Honeycomb?We were built for this.
The bottleneck in software development has shifted to understanding and validating what you just shipped. Everyone building AI observability has the same requirements for the data store powering it. Honeycomb has had this under the hood for years.Get a demoStart for freeA NEW ERA OF SOFTWARE
[H2] Speed without visibility is a liability
The cost of generating code has fallen close to zero, and the old ways of monitoring won't scale. Agentic systems demand massive throughput, access to context through high-cardinality data, and the ability to query what's happening right now. That's Honeycomb.
[IMG: Traditional APM Diagram]
[H3] Traditional monitoring won’t scale with AI
Most enterprise observability platforms use consumption or ingest-based pricing, where preserving more data directly increases cost. AI-powered systems generate high-cardinality, high-volume telemetry at a scale that forces a painful tradeoff: control costs by discarding data, or retain the context you need to explain system behavior and watch your bill explode. Without that data, teams and agents are left debugging symptoms instead of causes.
[IMG: Use observability to inform decisions - 42%, Monitor performance of AI systems - 22%]
[H3] Most teams don't have enough visibility
The result is an industry that is chronically underinvested in observability — and increasingly exposed.Only 42% of engineering teams regularly use observability data to inform decisions.For AI systems, the gap is even wider: only 22% of organizations using or experimenting with AI are actively monitoring the performance of AI models.(Source: LeadDev Engineering Performance Report, 2025)THE HONEYCOMB DIFFERENCE
[H2] Observability built for thedemands of AI
Honeycomb was built ten years ago from first principles, by asking what software engineers actually need to understand their systems in production. As it turns out, coding agents need exactly the same thing: fast, high-context, exploratory access to what's happening in production.
[IMG: Built for the unpredictable]
[H3] Built for the unpredictable
Understanding what went wrong in a complex AI system requires the ability to ask novel, open-ended questions of your production data. Honeycomb's unified data model retains the high-value relationships across production data so your team can actually investigate, rather than spending time correlating signals across multiple platforms. Or better yet, ask Canvas to investigate for you.
[IMG: Built for speed]
[H3] Built for speed
Honeycomb's purpose-built columnar data store delivers sub-second query times even across high-cardinality, high-dimensionality data. With Honeycomb, AI has near-instant access to terabytes of data to analyze faster than humans ever could. That speed changes how your team works. Engineers can shift from passively monitoring dashboards to actively exploring what agent-built code is doing in production.
[IMG: Built for scale]
[H3] Built for scale
Eliminate redundancies in your observability tooling, and pay only once to store all the events you need in Honeycomb without being penalized for adding context. As you build more (and faster) with AI, rely on Honeycomb’s scalable performance and OpenTelemetry standards to efficiently and quickly deliver answers to teams and agents.
[H2] Bottom line benefits
Teams using Honeycomb report measurable impact across the metrics that matter most to engineering leaders and their organizations.See the study79%Faster to respond to and remediate performance issues45%Reduced number of unplanned outages42%Higher DevOps team productivity reported
[H3] What our customers say
Intercom
[H3] Observing LLM performance with traces
Intercom's Fin.ai is one of the most complex AI agents in production today. When an optimization to speed Fin up inadvertently created wasted LLM calls that Finance caught in a quarterly review, the team used traces rich with context to identify, fix, and verify the issue in real time. Then they put an SLO on it.SLO-based monitoring has proven superior for LLM reliability than traditional metrics.Given AI's unpredictable nature, SLOs help us catch and investigate anomalies without triggering a flood of noisy alerts.
[IMG: Intercom]
Kesha MykhailovStaff Engineer, Intercom60%Reduction in mean time to first tokenRealtime Cost Optimization with immediate visibility into LLM efficiencyRead Case StudyScribe
[H3] Investigation to root cause in five minutes
After moving to Honeycomb, root cause identification dropped from one hour to around five minutes. Full frontend-to-backend tracing with OpenTelemetry gave engineers a single view across their entire stack. And observability costs dropped by 75% compared to their previous solution.The culture shift was just as significant. Before Honeycomb, observability was an afterthought. Today, Scribe's engineers are integrating Honeycomb's MCP with Claude Code to investigate alerts and query system behavior directly from their development environment.Once engineers experience good observability with Honeycomb, there's no going back.Instrumentation ships with the code now. They won't have it any other way.
[IMG: Scribe]
Steven TanSenior DevOps Engineer, Scribe1h → 5 minaverage time to identify root cause during incidents75%reduction in observability costs90%full feature adoption across engineering in just monthsRead Case StudyHomeaglow
[H2] Improving slow job performance with MCP
Homeaglow's engineering team is small, fast-moving, and deeply invested in making every engineer as effective as possible. When they integrated Honeycomb's MCP with their AI coding tools, the impact was immediate: engineers could surface slow traces, identify bottlenecks like N+1 queries, and ship fixes in a fraction of the time.Since we started using MCP, we have been able to improve performance of our slowest jobs in hours instead of days.Engineers pair slow traces with our backend codebase to quickly identify N+1 queries, slow operations, or other bottlenecks and then work with tools such as Claude to build tests, then make improvements.
[IMG: Homeaglow]
James BaxleyVP of Engineering, Homeaglow0internally-caused incidents for 12+ months40xperformance gain in core booking flowRead Case StudyTrusted by innovators and enterprises worldwide
[IMG: Slack]
[IMG: Intercom]
[IMG: Dropbox]
[IMG: Booking.com]
[IMG: Heroku]
[IMG: duolingo]
[IMG: Hello Fresh]
[IMG: jack henry]
[IMG: Slack]
[IMG: Intercom]
[IMG: Dropbox]
[IMG: Booking.com]
[IMG: Heroku]
[IMG: duolingo]
[IMG: Hello Fresh]
[IMG: jack henry]
[H2] We’ve done this before
Observability transformation doesn’t happen overnight. We know; we’ve helped hundreds of engineering teams do it. Whether you’re starting from scratch or replacing a stack of fragmented tools, Honeycomb meets you where you are.Let's Chat
[H3] Start with quick wins
Honeycomb's team will help you find high-impact improvements fast, without requiring a full instrumentation overhaul on day one.

[H3] Go deep on what matters
Pick a high-impact service, workflow, or team. We'll be your partner to drive measurable results, whether that’s faster incident response, better developer feedback loops, more confident deploys.
[H3] Scale across the org
From one team to your entire engineering organization. We'll help you build the observability culture and practices that make it stick.

[H2] Keep learning…
[H3] Platform Overview
Discover the power of our purpose-built data store and query engine.
[H3] Pricing
Straightforward plans with no penalties for curiosity.

[H3] Case Studies
Hear from the engineering teams already making it work.
[H2] The door is open
AI has made good observability more important and more achievable than ever. Your software won't wait.Get a demoStart for free

[H1] Learn observability from the experts
Our systems are exploding in complexity. Observability, once regarded as a nice-to-have, is now a crucial tool engineering organizations must leverage in order to debug quickly and deliver a world-class customer experience.Let authors Charity Majors, Liz Fong-Jones, and George Miranda show you what observability is and how to apply it successfully in your organization. Don’t wait—download your copy today.Get your copy
[IMG: Observability Engineering]
[H2] Download your copy of Observability Engineering
The term “observability” has expanded from the corner of systems engineering to the foundation of software, SaaS, and AI development. As observability has gained notoriety, it has been confused with adjacent concepts like monitoring, visibility, and telemetry.The book bridges both the theoretical and practical to provide examples of structured events to demonstrate the building blocks necessary for observability. Follow its lineage through core concepts like distributed tracing, iterative verification of hypotheses, and debugging from first principles with the core analysis loop.
[H6] Get your complimentary copy
[H2] What you’ll learn in the book
[H3] See clearly, build better
Understand the value of observability when managing complex cloud-native apps and systems.
[H3] Clarity for every engineer
See the impact observability has across the entire software engineering cycle.
[H3] Teamwork makes the SLOs work
Learn how different teams can work together to draft SLOs that work for both business and engineering.
[IMG: Happy Face]
[H3] From code to customer impact
Recognize how software developers contribute to the customer experience and business impact.
[H3] Debug with context
Produce quality code for context-aware debugging and maintenance.
[IMG: Lightning Bolt]
[H3] Faster fixes with better telemetry
Leverage data-rich analytics to find answers quickly when maintaining site reliability.
[H2] What's next for Observability Engineering?
[IMG: Observability Diagrams]
[H2] Second edition, coming in 2026
Three years ago, when we wrote Observability Engineering, the landscape was quite different. We’re proud to announce that the second edition of the book is coming in 2026, with 32 new chapters that cover issues like cost, governance, AI, and more.Sign up for updates
[H2] Meet the authors of Observability Engineering
[IMG: Charity Majors]
[H3] Charity Majors
Co-founder & CTO of HoneycombCharity pioneered the concept of modern observability, drawing on her years of experience building and managing massive distributed systems at Parse, Facebook, and Linden Lab.She is the co-author of Observability Engineering and Database Reliability Engineering (O’Reilly).She loves free speech, free software, and single malt scotch.
[IMG: Liz Fong-Jones]
[H3] Liz Fong-Jones
Field CTO, HoneycombLiz is a developer advocate, labor and ethics organizer, and Site Reliability Engineer (SRE) with over two decades of experience. She is currently the Field CTO at Honeycomb, and previously was an SRE working on products ranging from the Google Cloud Load Balancer to Google Flights.She lives in Vancouver, BC with her wife Elly, partners, and a Samoyed/Golden Retriever mix, and in Sydney, NSW. She plays classical piano, leads an EVE Online alliance, and advocates for transgender rights.
[IMG: George Miranda]
[H3] George Miranda
VP Marketing, InsightFinder AIGeorge is a former systems engineer now working to promote developer tools he believes in. Previously, he spent more than 15 years building large-scale distributed systems in the finance and video games industries.In his free time, George likes to do things far less dangerous than getting into arguments about observability—like motorcycle racing and helicopter snowboarding.
[H2] Ready to get started?
Get the bookStart for free