Datadog Bits AI SRE Alternative: Open Source, Self-Hosted, Vendor-Neutral

Key Takeaways

• Datadog Bits AI SRE went generally available on December 2, 2025, Datadog's first generally available AI agent, after testing across more than 2,000 customer environments.
• Bits reasons over the full breadth and depth of the Datadog platform's data. Cross-platform reasoning over GitHub, ServiceNow, Grafana, Splunk, Dynatrace, and Sentry is in Preview, not GA.
• Bits is priced through Datadog's AI Credits model and billed per conclusive investigation, stacking on top of existing host, APM, logs, and RUM licensing. Datadog does not publish a flat per-investigation dollar rate.
• Aurora is open source under Apache 2.0, self-hosted, air-gapped capable, and bring-your-own-LLM with Ollama.
• Aurora investigates across AWS, Azure, GCP, OVH, Scaleway, and Kubernetes and ingests alerts from 11 monitoring connectors, including Datadog itself, so you keep Datadog as a data source without locking your reasoning layer to it.

Datadog Bits AI SRE is one of the strongest agentic incident investigators shipping today, and if your stack is already all-in on Datadog it is a natural fit. This post is an honest look at where Bits excels, where its single-vendor design and billing model become friction, and how the open source Aurora project offers a self-hosted, vendor-neutral alternative for teams that want auditability and want to avoid stacking another usage meter on the bill.

What is Datadog Bits AI SRE?

Bits AI SRE is Datadog's autonomous incident investigation agent, generally available since December 2, 2025. When an alert fires, Bits automatically analyzes runbooks, telemetry, and more to separate signal from noise and surface hypothetical root causes, then delivers conclusions into collaboration tools before an on-call engineer even logs in.

It is Datadog's first generally available AI agent, and it earned that status the hard way. Datadog says it tested Bits across more than 2,000 customer environments and ran tens of thousands of investigations before GA. The newer agent harness lets Bits plan investigations, evaluate competing hypotheses, and refine in real time, running roughly twice as fast as before in about 3 to 4 minutes depending on complexity.

Bits has real strengths worth naming. It is deeply integrated with the Datadog platform, it supports HIPAA-regulated workloads with role-based access controls, and it learns from each investigation. For a team that already pays for Datadog infrastructure monitoring, APM, logs, and RUM, Bits reads that telemetry with zero additional plumbing.

What is Aurora?

Aurora is an open source, Apache-2.0 AI SRE and incident-management platform built by Arvo AI. Like Bits, it autonomously investigates incidents and produces root-cause analyses, but it is designed to be self-hosted and vendor-neutral rather than tied to one observability backend.

Aurora's LangGraph-orchestrated agents query AWS, Azure, GCP, OVH, Scaleway, and Kubernetes, then run kubectl, aws, az, and gcloud commands inside sandboxed Kubernetes pods to gather live evidence. It builds a Memgraph infrastructure knowledge graph to reason about blast radius, generates root-cause analyses and postmortems that export to Confluence, Notion, or SharePoint, and can suggest code fixes or open pull requests. Destructive actions are human-gated. Because it is bring-your-own-LLM with Ollama, Aurora can run fully air-gapped with no telemetry leaving your environment.

Aurora ingests alerts via webhook from 11 monitoring connectors: PagerDuty, Datadog, Grafana, New Relic, OpsGenie, Netdata, Dynatrace, Coroot, ThousandEyes, BigPanda, and incident.io, plus a Slack bot. Notably, Datadog is one of those connectors, so Aurora can keep Datadog as a telemetry source while running the reasoning layer on infrastructure you control.

Bits AI SRE vs Aurora: the head-to-head

The core difference is scope and control. Bits is a closed, SaaS-only agent that reasons over Datadog-collected data and meters each investigation, while Aurora is an open source, self-hosted agent that reasons across multiple clouds and monitoring tools with no per-investigation meter.

Where Bits reasons, and where it does not yet

At GA, Bits reasons over the full breadth and depth of the Datadog platform's data: metrics, logs, traces, dashboards, changes, source code, RUM, Database Monitoring, Network Path, and Continuous Profiler. That is a lot of context, but it is Datadog context. Pulling telemetry from outside tools such as GitHub, ServiceNow, Grafana, Splunk, Dynatrace, and Sentry is currently a Preview feature you have to sign up for, not a GA capability. If a meaningful slice of your signal lives outside Datadog, Bits cannot reason over it today without that preview access. Aurora ingests from 11 monitoring connectors out of the box and queries cloud and Kubernetes APIs directly, so its evidence is not gated on a single vendor's collection pipeline. See our guide to multi-cloud incident management for why that matters in mixed-cloud environments.

Investigation versus execution

Both tools genuinely investigate rather than just correlating alerts, which is the right bar in 2026. Bits delivers conclusions and supports a set of triage actions, including sending Slack and Microsoft Teams messages, creating incidents, paging engineers, creating cases, and filing Jira tickets, per Datadog's deeper-reasoning writeup. Aurora goes a step further on the execution side by running read-only CLI commands in sandboxed pods to gather evidence and, with human approval, suggesting code fixes or opening pull requests. Both keep destructive actions behind a human gate.

Deployment and auditability

Bits runs only in Datadog's SaaS. There is no on-premises or air-gapped option, and the reasoning model is closed. Aurora is self-hosted via Docker Compose or a Helm chart, can run air-gapped with local models, and is fully open source, so security teams can read every line of the agent that touches production. For regulated or sovereignty-constrained teams, that audit surface is often the deciding factor, as we cover in self-hosted AI SRE.

Comparison table

Dimension	Datadog Bits AI SRE	Aurora
License	Proprietary, closed source	Open source, Apache 2.0
Deployment	Datadog SaaS only	Self-hosted, air-gapped capable
Multi-cloud reasoning	Datadog-collected data; cross-tool reasoning in Preview	Direct queries to AWS, Azure, GCP, OVH, Scaleway, Kubernetes
Investigation vs correlation	Plans and tests hypotheses over Datadog telemetry	Plans and tests hypotheses; builds Memgraph blast-radius graph
Write and execute actions	Triage actions: Slack, Teams, incidents, paging, cases, Jira	Sandboxed read-only CLI; human-gated PRs and fixes
LLM choice	Datadog-managed models	Bring-your-own-LLM: Ollama
Pricing model	AI Credits, billed per conclusive investigation on top of existing licensing	Free, no per-seat or per-investigation fee
Monitoring inputs	Datadog platform; outside tools in Preview	11 webhook connectors, including Datadog itself

How is Bits AI SRE priced?

Bits AI SRE is sold through Datadog's AI Credits model and billed per conclusive investigation, and that charge stacks on top of whatever you already pay for hosts, APM, logs, and RUM. Datadog does not publish a flat per-investigation dollar rate; the cost is expressed in credits, and only investigations that reach a conclusive status are billable, so spend rises with alert and incident volume.

That model has two practical consequences. First, it adds a new usage meter on top of an already usage-metered bill, which can make a busy incident week more expensive in the exact moment your team is under pressure. Second, because the per-investigation cost is expressed in credits rather than a single posted price, forecasting requires modeling your own conclusive-investigation volume. We are deliberately not quoting a per-investigation dollar figure here, because Datadog does not publish one; treat any specific number you see elsewhere as a third-party estimate, not an official rate.

Aurora's pricing model is the opposite by design: it is free and open source with no per-seat or per-incident pricing. Your only costs are the infrastructure you run it on and whatever LLM you point it at, which you choose and control. If you run a local model via Ollama, inference cost can be effectively fixed.

Where does Bits AI SRE fit against the rest of the market?

Bits is one entry in a fast-moving category, and it is not the only credible option. Closed SaaS investigators like Resolve.ai and the AI features inside PagerDuty and Dynatrace compete for the same incident, and AIOps correlation engines such as Dell APEX AIOps Incident Management, the product formerly known as Moogsoft, still anchor a lot of enterprise NOCs with ML-based correlation. Dell APEX is actively maintained, not discontinued, but it is proprietary and enterprise-licensed, and correlation is a different job than agentic investigation.

It is also worth being precise about routing. Alert routing, scheduling, and escalation tools are complementary to an investigation agent, not substitutes for one. Grafana archived the open source grafana/oncall repository on March 24, 2026 and pointed users toward Grafana Cloud IRM, so teams that relied on OSS OnCall are migrating their routing layer. Aurora does not replace that routing layer; it sits on top of whatever you move to, whether that is an open source router like Keep or notifications via ntfy or Twilio, and adds the AI investigation step. For the broader landscape, see our roundup of the top AI SRE tools in 2026.

Which should you choose?

Choose Bits AI SRE if your organization is committed to Datadog as its single observability platform and you want the lowest-friction agent that reasons over that data, with HIPAA support and RBAC, and you are comfortable adding a per-investigation meter to your Datadog bill. The integration depth is real, and for an all-Datadog shop it is hard to beat on setup effort.

Choose Aurora if you want a vendor-neutral, self-hosted agent that reasons across multiple clouds and monitoring tools, keeps Datadog as one input rather than the only one, runs with the LLM you choose, can operate air-gapped, and costs nothing per investigation. The open source license also means you can audit exactly what the agent does before it touches production. If you are weighing several options, our framework for how to evaluate an AI SRE platform walks through the criteria that matter most.

The two are not mutually exclusive. A common pattern is to keep Datadog for telemetry, wire its alerts into Aurora's Datadog connector, and let an open, auditable agent do the cross-cloud investigation on infrastructure you control.

You can read the code, deployment guides, and connector list in the Aurora repository on GitHub.