NeuBird
Also known as: NeuBird Hawkeye, NeuBird Falcon
Agentic reliability platform whose AI SRE agents investigate production alerts across 50+ tools in place, with approval gated write access, customer governed model routing and an MCP server.
NeuBird sells what it calls an Agentic Reliability Center: one governed place for the agents that touch production. Its agents triage alerts arriving from PagerDuty and Slack, investigate by querying metrics, events, logs and traces in place across more than 50 integrations without storing the telemetry, and report root cause with a chain of thought, the queries run and the sources consulted. Conclusions from each investigation and approved fix are kept as institutional memory and surfaced when related symptoms recur.
Write access is earned in tiers, Suggest, Recommend and Act, and every action waits for a human approval recorded on the audit trail, with autonomy configurable per environment and role based permissions limiting what each agent may reach. Governed Model Access lets the customer choose approved models, including ones in its own AWS Bedrock, Azure OpenAI or Google Vertex AI account, route by task, cost or latency, and cap spend per agent. The platform runs as SaaS or inside the customer's VPC, and an MCP server with 33 tools lets coding and ops assistants create projects, start investigations and test instructions against past sessions with an RCA quality score.
Pricing is an Enterprise plan metered in credits per agent run, with triage included, alerts unlimited and the credit price quoted by sales. A 14 day trial gives 50 credits. NeuBird holds SOC 2 Type II.
Vendor details
Canonical URL
https://neubird.ai
Category
SRE / DevOps agent
Subcategory
AI SRE agent
Funding status
Independent, closed a $19.3 million round led by Xora Innovation in April 2026, bringing total funding to $63.8 million, with Mayfield, StepStone Group, Prosperity7 Ventures, and M12 (Microsoft's venture fund). Co founders Gou Rao and Vinod Jayaraman previously built Portworx, acquired by Pure Storage, and Ocarina Networks, acquired by Dell.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Queries 50+ tools in place, including Datadog, AWS, Azure, Google Cloud, PagerDuty, Dynatrace, Splunk, Snowflake, Grafana, OpenShift, GitHub and GitLab, and works with Slack, Microsoft Teams, Jira and ServiceNow. Exposed to assistants such as Claude Code, Cursor and GitHub Copilot through an MCP server, plus a terminal CLI and a desktop app.
In practice
Security will not let telemetry leave the environment. NeuBird queries metrics, logs and traces in place with zero telemetry storage and can run inside your VPC.
You want agents to fix production but only with sign off. NeuBird's Earned Write Access moves from suggest to act, and every action waits for a recorded human approval.
Model choice is a governance question. NeuBird routes agents only to approved models, including ones in your own Bedrock, Azure OpenAI or Vertex account, with per agent budgets.
Sources & related URLs
Agentic Index coverage score
11.0 / 14 capabilities · 79%
| Integrations & Tool Calling | Full |
|---|---|
|
NeuBird queries more than 50 named integrations in place, including Datadog, AWS, Azure, Google Cloud, PagerDuty, Dynatrace, Splunk, Snowflake, Grafana, GitHub, GitLab, OpenShift, MongoDB and Redis, with Jira and ServiceNow incident workflows. Connections are created and scoped per project. Sourceneubird.airead 2026-09-29 |
|
| Workflow Orchestration | Partial |
|
Triage, investigation, analyst, cluster and discovery agents run as stages of NeuBird's own investigation pipeline, correlating evidence across integrations in parallel. No builder, branching or agent coordination that customers control is described. Sourceneubird.ai/pricingread 2026-09-29 |
|
| Knowledge Grounding & RAG | Partial |
|
The agent queries raw metrics, events, logs and traces directly across all of the customer's observability tools, with no telemetry stored, so context is assembled for each run. No maintained index over the customer's knowledge is described. Sourcedocs.neubird.ai/concepts/overviewread 2026-09-29 |
|
| Human Oversight & Guardrails | Full |
|
Under Earned Write Access, every action waits for a human approval, and the audit trail shows who gave it. Autonomy moves through Suggest, Recommend and Act tiers, high risk actions need human confirmation before they run, and autonomy levels can be set per environment. Sourceneubird.ai/securityread 2026-09-29 |
|
| Security, Identity & Governance | Full |
|
Role based access control gives agents access only to the systems and actions explicitly granted to them, and agents never hold credentials. NeuBird cites SOC 2 Type II, and a trust center sits at trust.neubird.ai. SSO and SCIM support is not stated. Sourceneubird.ai/securityread 2026-09-29 |
|
| Observability & Auditability | Full |
|
Every observation, inference and action is logged with timestamps, context and outcomes, and each investigation exposes its chain of thought, query execution logs and the data sources consulted. Model calls carry a per call audit of agent, task, model, tokens and cost. Sourcedocs.neubird.ai/mcp/reference/toolsread 2026-09-29 |
|
| Memory & State Persistence | Partial |
|
Every investigation and approved fix becomes institutional memory, kept as conclusions with no telemetry stored, and surfaces automatically on related symptoms. No scope, lifetime or delete path for that memory is published. Sourceneubird.airead 2026-09-29 |
|
| Deployment & Data Residency | Full |
|
Enterprise offers SaaS or VPC/VNET deployment, NeuBird can deploy within the customer's cloud environment, and resources can sit on-prem or in the customer's VPC, with telemetry staying in place and nothing stored. No hosting region list is published. Sourceneubird.ai/pricingread 2026-09-29 |
|
| Prebuilt Agents, Templates & Packs | Partial |
|
Pricing is per run for the triage, investigation, analyst, cluster and discovery agents, but they work as stages of one investigation product, not agents a customer adopts separately. Customers write their own project instructions, and no template or agent catalog is published. Sourceneubird.ai/pricingread 2026-09-29 |
|
| Triggers & Channel Coverage | Full |
|
PagerDuty alerts come in and are triaged by the included Triage Agent, with unlimited alerts always included, and Slack channels surface production issues. Engineers also start investigations from Slack, the desktop app, the terminal or MCP. Sourceneubird.ai/pricingread 2026-09-29 |
|
| Model Flexibility & Routing | Full |
|
With Governed Model Access, customers decide which models run, on which data and at what cost, using approved frontier models, including ones in their own AWS Bedrock, Azure OpenAI or Google Vertex AI account. Requests are routed by task, cost or latency, with failover and per agent budgets. Sourceneubird.ai/platform/governed-model-accessread 2026-09-29 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
A remote MCP server at <deployment>.app.neubird.ai/mcp, also available as the @neubirdai/mcp-server-neubird package, authenticates with a bearer token from the login API and exposes 33 tools, including ones that create projects, connections and instructions and start investigations. Sourcedocs.neubird.ai/mcp/getting-started/installationread 2026-09-29 |
|
| Testing, Debugging & Optimization | Full |
|
A customer can test a change to its configuration by validating an instruction, applying it to a past session, running the investigation again and reading the RCA quality score and session scoring, through the neubird_validate_instruction, neubird_apply_session_instruction, neubird_rerun_session and neubird_get_rca_score tools. Sourcedocs.neubird.ai/mcp/reference/toolsread 2026-09-29 |
|
| Browser & Computer Use | Not documented |
|
NeuBird works through telemetry queries, integrations, a terminal, a desktop app and an MCP server. No browser, desktop control or computer use by the agent is described. Sourcedocs.neubird.ai/concepts/overviewread 2026-09-29 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Product updates
Version 5.0.0 of the NeuBird MCP server turns off by default the nine tools that create, change or delete a tenant's configuration, covering connections, projects and skills. They are not listed, and a direct call is refused unless NEUBIRD_ENABLE_CONFIG_TOOLS is set to true. Every tool now carries MCP annotations that mark it read only or destructive, so clients can run read only tools without asking and prompt before destructive ones.
Bears on: MCP / tool calling / API
View sourceNeuBird AI released a new native integration with ControlUp to bring automated root cause analysis to virtual desktop infrastructure (VDI) monitoring. The integration acts as an agentic analyst that continuously analyzes performance metrics, identifies relationships between hosts and sessions, and automates VDI investigations.
Bears on: Integrations
View sourcePricing
Enterprise at custom pricing on credits. A 14 day trial includes 50 credits.
Credits per agent run. Investigation, analyst and cluster runs draw credits, triage is included, discovery is charged per 10 runs, and alerts are unlimited.
Included quota
The trial includes 50 credits, up to 5 users and 1 project for 14 days. Enterprise includes unlimited alerts, users and projects, and unlimited historical storage.
What is public
NeuBird publishes the credit cost of each agent run, its credit guidance by alert volume (about 100 credits a month for 1,000 alerts, about 1,000 for 10,000), the Enterprise inclusions and the trial terms. The price of a credit is not published.
Billing mechanics
Custom Enterprise contracts are metered in credits, consumed when agents investigate, analyze or answer questions. Unlimited alerts, users and projects, SaaS or VPC deployment, MCP access and 24/7 support are included.
Cost watchouts
Credit consumption grows with the alert volume that turns into investigations. Alert volumes above 50,000 a month move to custom enterprise pricing.
Variable cost rationale
Credits are metered per agent run: 1 for an investigation or analyst run, 2 for a cluster run and 1 per 10 discovery runs, with triage included and alerts unlimited. The bill grows with the number of alerts that turn into investigations, about 100 credits a month for 1,000 alerts and about 1,000 for 10,000. The credit price is set in a custom Enterprise contract. No cap or overage rate is published.
Additional watchouts
Spend scales with the number of investigations and cluster runs rather than seats, and the credit price comes only from sales.
Overage / add-ons
Not published. Credit prices are custom.
Sales call required
Yes, required for paid access
Free / trial
14 day trial with 50 credits, up to 5 users and 1 project, no credit card required
Lowest paid plan
Not published, Enterprise is on custom credit pricing
Commercial notes
Independent, with $63.8 million raised in total after a $19.3 million round led by Xora in April 2026, and backed by Microsoft's M12. The founders built Portworx (Pure Storage) and Ocarina (Dell). NeuBird cites more than one million alerts resolved and MTTR reductions of 88 to 90 percent.
Key ambiguities
The credit meter and the per run costs are published. The price per credit is not.
Support SLA / resale
Enterprise includes 24/7 support, with SaaS or VPC deployment.
Related vendors
- AlertD — AI agents for AWS operations that run read only in the customer's…
- Anyshift — AI SRE built on a versioned knowledge graph of infrastructure, apps…
- Avesha — Obliq, Avesha's autonomous AI SRE for Kubernetes and agentic…
- Better Stack — Better Stack offers incident management with built in on call,…
- Bluebricks — Context and control layer that lets AI agents operate cloud…
- Cased — AI workflows for infrastructure and platform engineers: default and…
Alternatives to NeuBird
The closest documented capability profiles to NeuBird among SRE and DevOps agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Chamber9.5 / 14A lighter documented profile than NeuBird
- Doctor Droid11.5 / 14Fuller documented coverage on Workflow Orchestration and Knowledge Grounding & RAG
- Fabrix.ai12.5 / 14Fuller documented coverage on Workflow Orchestration and Knowledge Grounding & RAG
- Itential11.5 / 14Fuller documented coverage on Workflow Orchestration and Prebuilt Agents, Templates & Packs
- Cleric10.0 / 14Fuller documented coverage on Memory & State PersistenceNeuBird vs Cleric →
- Komodor13.0 / 14Fuller documented coverage on Workflow Orchestration and Knowledge Grounding & RAG
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded