Datadog tells you something is wrong. Struct tells you where, why, and what to do about it.
Datadog Bits AI SRE and Struct both auto-investigate production alerts - neither waits for an engineer to start typing. The difference is scope and where the engineer works: Bits AI investigates inside Datadog telemetry, while Struct is a persistent, conversational on-call agent that investigates across your entire production stack — Sentry, Datadog, GitHub, logs, traces, and code — dedupes the noise automatically, and posts a cited root-cause hypothesis to Slack as it works.
Does Struct work with Datadog?
Datadog is an observability platform: engineering teams use it for infrastructure monitoring, APM, log management, dashboards, and alerting across their production environment. It's where your metrics, logs, and traces already live. Struct doesn't replace that: it's a complementary agentic layer that ties this and other sources and signals together.
Struct connects to your Datadog metrics, logs, and traces and uses them as primary inputs. It adds cross-stack investigation that reaches beyond Datadog telemetry into Sentry stack traces, GitHub commit history, Cloud logging, and any other tool an on-call engineer would normally open mid-incident. Most Struct customers keep their Datadog deployment and add Struct on top.
What's the same between Struct and Datadog Bits AI?
Both tools are automated. Neither waits for an engineer to start typing. Both kick off an investigation when something fires, and in practice both take roughly the same amount of time to reach a diagnosis. The difference engineers feel first is where and how fast the signal shows up: Struct starts streaming its findings into Slack while it works, so the on-call engineer watches the hypothesis take shape in real time instead of waiting for a finished report inside another tool.
Where do Struct and Datadog Bits AI differ?
Dimension | Struct | Datadog Bits AI SRE |
|---|---|---|
Investigation surface | Cross-stack: Sentry + Datadog + GitHub + logs + traces + code, correlated in one report. | Primarily Datadog telemetry — metrics, logs, traces, monitors. External connections are alpha. |
What triggers it | Automated across the stack — an alert or mention from anywhere your engineers work can kick off an investigation. | Automated, but triggered primarily by Datadog and operating on Datadog data. |
Noise & deduping | Auto-dedupes related alerts with no configuration. One incident, one investigation. | No auto-deduping — you tune your own alerts/monitors or configure manual keyword filters to cut noise. |
Where you work it | Slack-native and conversational. Streams updates as it investigates, answers follow-ups in-thread. | Datadog-UI-focused. Slack integration exists but is lighter. |
Incident workflows | Auto-integrates with incident-channel tooling like incident.io and Rootly. | No automatic integration with incident-channel workflows. |
Memory | Persistent agent that can run investigations - knows current state, remembers prior incidents, systems, and past investigations. | Session-oriented, with ability to see current/past state and remember feedback. |
Pricing | Credit-based - base fee includes full platform access and discounted credits. Auto-deduping keeps investigation volume and cost predictable. | Credit-based, and more expensive in practice, especially without deduping, since every undeduped alert burns credits. Variable pricing for different use cases. |
Deduping
Struct: Correlates related alerts automatically and runs one investigation per real incident — no config, no keyword lists to maintain. Less noise for the on-call engineer and fewer redundant investigations to pay for.
Datadog Bits AI SRE: No auto-deduping. To keep an alert storm from turning into a flood of separate investigations, you tune your monitors and alerts yourself or set up manual keyword filtering.
Slack depth
Struct: Slack is a first-class surface. The investigation posts into the on-call channel, streams its reasoning as it goes, and stays conversational. The engineer can ask follow-up questions in-thread and get answers without leaving Slack.
Datadog Bits AI SRE: Built around the Datadog UI. There's a Slack integration, but it's not as deep or conversational - the engineer generally ends up in Datadog to do real work.
Incident workflows & memory
Struct: Plugs directly into incident-channel workflows like incident.io and Rootly, and carries memory across incidents — it remembers your systems and what past investigations found, so each one starts smarter than the last.
Datadog Bits AI SRE: No automatic incident.io/Rootly channel integration, and operates as a session-oriented in-product assistant rather than a persistent agent.
Cost
Struct: Pricing stays predictable because deduping holds down the number of investigations that actually run.
Datadog Bits AI SRE: Credit-based pricing that runs more expensive in practice — and without deduping, every redundant alert spends credits.
Where Datadog Bits AI is strong
Credit where it's due: Bits AI SRE is deeply integrated into Datadog, and the native in-app experience is excellent. Each investigation gets a clean visualizer right alongside the telemetry it's reasoning over. If your investigations rarely cross into Sentry, GitHub, or other tools, and your engineers already live in Datadog, Bits delivers a strong experience without adding a new vendor.
What does Struct + Datadog look like in production?
Arcana (Series B fintech, 40 engineers) ran their investigations manually across Sentry + Datadog. They added Struct on top of their existing Datadog deployment:
Datadog for metrics, traces, dashboards, alerting
Struct for automatic, deduped investigation on every alert
Numbers from their deployment, against their own manual baseline:
Metric | Before Struct (manual, Sentry + Datadog) | After Struct |
|---|---|---|
Median investigation time | 30 min | 2 min |
Senior eng hours/month on investigation | ~60 | ~4 |
Investigations / month | ~150 (manual cap) | 2,100+ |
Helpful rate | n/a | >80% |
Same Datadog spend. 56 hours/month of senior engineer time back. Arcana's CTO is on record saying Struct gave their senior engineers their time back.
Bottom line: keep Datadog, add Struct on top
Both Struct and Bits AI SRE auto-investigate, and both take about the same time to reach an answer. The outcome for most teams is the same too: they keep Datadog and add Struct on top — for cross-stack investigations, a persistent and conversational on-call agent with memory, auto-deduping with no config, native incident.io/Rootly workflows, and a more cost-effective model. Bits AI SRE remains a strong native assistant for teams whose alerts and investigations stay inside Datadog.
Information about Datadog Bits AI SRE is based on publicly available data as of June 2026 and may not reflect the most current features or pricing. All third-party trademarks referenced on this page are the property of their respective owners. Struct is not affiliated with, endorsed by, or sponsored by Datadog.
