Cleric
AI SRE that verifies each change after merge against live production, investigates alerts read only, and hands fix pull requests to engineers, with memories that carry context between investigations.
Cleric is an AI SRE that follows every change after merge, checks its effect against live production metrics, triages and investigates alerts, and hands off fixes as pull requests for engineers to review. Its access is read only, enforced by role based permissions and per integration scopes. Cleric does not make changes to customer infrastructure, and all remediation is performed by the customer's team. Work reaches it from alert bots posting in Slack, from change verification after a merge, and from scheduled agents, and customers can add their own agents written as a prompt beside the built in investigator.
Each investigation keeps an activity log of the plan, the tools used, the exact commands and queries run, their raw output and the reasoning behind the conclusion. Memories, facts Cleric proposes from conversations or that engineers add in Slack, are organized by service and environment, searched automatically in later investigations, and can be edited or deleted; stale or contradicted memories are cleaned up. Documentation from Atlassian and GitHub is searched during investigations, and message ratings and corrections feed the agent's calibration. An MCP server lets other agents start and read investigations and search Cleric's knowledge base.
Pricing is published on credits: Starter $100 a month for 100 credits, Team $600, Pro $2,000 and a custom Enterprise pool, with an investigation or change verification costing 10 credits and a 500 credit evaluation. Enterprise SSO, the organization audit log and scheduled agents start at Pro, and EU hosting and custom region at Enterprise. The service holds SOC 2 Type II and names Anthropic, Google Gemini and OpenAI as its model providers.
Vendor details
Canonical URL
https://cleric.ai
Category
SRE / DevOps agent
Subcategory
AI SRE agent
Funding status
Independent, raised a $4.3 million seed round led by Zetta Venture Partners. Named a Gartner Cool Vendor 2025 in AI for SRE and Observability. Publishes headline metrics of five minutes to root cause, 92 percent actionable findings, and more than 200,000 production grade investigations completed.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Connects read only to observability, infrastructure, code and documentation tools, including AWS, Azure, Datadog, Grafana, PagerDuty, Atlassian and GitHub, with access scoped per integration, and works in Slack and a web app. An MCP server exposes investigations and the knowledge base to other agents, and a private network connector reaches private resources on Pro.
In practice
You want autonomous investigation but will not give an agent write access to production. Cleric investigates read only and leaves remediation to your team.
A deploy goes out on Friday afternoon. Cleric checks the change against live production metrics after merge and investigates if it regresses.
Your on call engineer keeps rediscovering the same quirks. Cleric stores memories by service and environment and applies them in the next investigation.
Sources & related URLs
Agentic Index coverage score
10.0 / 14 capabilities · 71%
| Integrations & Tool Calling | Full |
|---|---|
|
Cleric has named integrations across observability, infrastructure, code and documentation (AWS, Azure, Datadog, Grafana, PagerDuty, Slack, Atlassian, GitHub), with scoped access for each. The agent runs its own commands and queries against them and opens pull requests with fixes. Sourcecleric.airead 2026-09-29 |
|
| Workflow Orchestration | Partial |
|
One investigating agent plans tasks, runs queries and adjusts its plan. Custom agents are a name plus a free text prompt fired by a trigger, with no steps, branching or handoffs between agents, and the built in investigate agent cannot be edited. Sourcedocs.cleric.ai/setup/agents-and-triggersread 2026-09-29 |
|
| Knowledge Grounding & RAG | Partial |
|
Documentation integrations (Atlassian, GitHub) let the agent search and retrieve documentation during investigations, and customers add Global Guidance in settings. Context is assembled for each run, and there is no maintained index over the customer's documents. Sourcedocs.cleric.ai/setup/providing-contextread 2026-09-29 |
|
| Human Oversight & Guardrails | Partial |
|
Access is read only, enforced by RBAC and access scopes. Cleric does not make changes to customer infrastructure but may propose them, and all remediation actions are performed by the customer's team. People keep the decisions, and the agent's reach is scoped. Fix pull requests are approved in the customer's own code host. The approve button in the homepage mockup is not explained as a Cleric approval step. Sourcedocs.cleric.ai/security/security-data-privacyread 2026-09-29 |
|
| Security, Identity & Governance | Full |
|
Customers control access through RBAC and access scopes for each integration, and code access can be limited to one repository. Every plan has Google Workspace sign in, and Pro adds Okta, custom SAML or OIDC and an organization audit log. Cleric holds SOC 2 Type II, has a Trust Center and runs annual penetration testing, with AES-256 encryption at rest. Sourcedocs.cleric.ai/security/security-data-privacyread 2026-09-29 |
|
| Observability & Auditability | Full |
|
Each investigation keeps an activity log recording the plan, the tools used, the specific commands and queries executed, the raw output from each tool and the reasoning behind conclusions, viewable in the web app. Retained investigation data includes queries, reasoning traces and cited evidence. Sourcedocs.cleric.ai/usage/reviewing-the-activity-logread 2026-09-29 |
|
| Memory & State Persistence | Full |
|
Memories are facts Cleric proposes from conversations or users add with '@Cleric Remember', organized by service and environment, searched automatically during investigations, kept until deleted or removed by automatic cleanup when stale or contradicted, and viewable, editable and deletable in the Knowledge section. Sourcedocs.cleric.ai/learning/memoriesread 2026-09-29 |
|
| Deployment & Data Residency | Full |
|
EU hosted instances run the application, database and model inference in Europe, beside the default GCP hosting, and Enterprise offers custom data retention and region. Operational logs, traces and product analytics are still processed centrally in the US. A private network connector reaches private resources but does not change where Cleric runs. Sourcedocs.cleric.ai/security/eu-hosting-data-processingread 2026-09-29 |
|
| Prebuilt Agents, Templates & Packs | Partial |
|
One built in investigate agent ships and cannot be edited. Further agents are ones the customer writes as a name and a prompt, and there is no template or agent catalog. Sourcedocs.cleric.ai/setup/agents-and-triggersread 2026-09-29 |
|
| Triggers & Channel Coverage | Full |
|
Agents fire on bot messages posted in Slack channels that match trigger keywords (alert bots such as PagerDuty and Datadog) and on recurring cron schedules, and change verification follows each change after merge. Scheduled agents and monitors are on the Pro plan. Sourcedocs.cleric.ai/setup/agents-and-triggersread 2026-09-29 |
|
| Model Flexibility & Routing | Partial |
|
Cleric names Anthropic, Google Gemini and OpenAI as its model providers, under zero training and zero retention terms. Cleric uses them internally, and customers and admins cannot choose a model. Sourcedocs.cleric.ai/security/security-data-privacyread 2026-09-29 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
Outside callers drive Cleric through its own MCP server, on every plan. It has a per tenant endpoint at <company>.app.cleric.ai/mcp, bearer authentication with personal API keys and an enumerated tool list, and its create_issue tool starts a new investigation, a core platform object. There is no REST API or SDK. Sourcedocs.cleric.ai/integrations/mcpread 2026-09-29 |
|
| Testing, Debugging & Optimization | Partial |
|
A learning loop calibrates the agent through explicit feedback (1 to 5 message ratings and corrections), implicit feedback from actions taken, and analysis of past investigations. There is no evaluation harness, scored test or quality metric for findings. Change verification tests the customer's releases, not the agent. Sourcedocs.cleric.ai/learning/how-cleric-learnsread 2026-09-29 |
|
| Browser & Computer Use | Not documented |
|
Work runs through scoped integrations, commands and queries, Slack and an MCP server, with no browser, desktop or computer control. Sourcedocs.cleric.ai/usage/reviewing-the-activity-logread 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
Cleric launched a production verifier that dynamically generates code to continuously poll live production telemetry and determine if an incident was successfully resolved. It evaluates whether the agent's suggested fix was actually applied and tracks if the underlying defect disappeared. These outcomes are then surfaced in a new in-product performance dashboard to account for the value delivered.
Bears on: Observability / auditability
View sourcePricing
From $700 per month on Team, with 1,400 credits. Pro is $2,000 and Enterprise is custom.
Monthly credit plans. A change verification or an investigation draws 10 credits and chat 1 credit per minute, and Enterprise runs on a custom annual credit pool.
Included quota
Team includes 1,400 credits a month, about 140 investigations or change verifications, and Pro 2,000.
What is public
Four plans carry published prices and credit allowances, per action credit costs and the features added at each tier. Team adds custom agents and alert routing. Pro adds scheduled agents and monitors, a private network connector, Okta and custom SAML or OIDC and an organization audit log. Enterprise adds custom data retention and region, a DPA and a security review.
Billing mechanics
A monthly subscription buys a credit allowance. Each change verification or investigation draws 10 credits, and chat draws 1 credit a minute. All plans include unlimited users, all integrations, MCP access, operational memory and investigation history.
Cost watchouts
Overage beyond the monthly credit allowance is not published, and SAML or OIDC SSO, the organization audit log and region choice sit on the $2,000 Pro plan or above.
Variable cost rationale
Cleric meters credits. Investigations and change verifications draw 10 credits each and chat 1 credit a minute, so the bill grows with how many changes and alerts the agent works and how long teams chat with it. The monthly plan allowance limits the bill, a busy estate moves up a plan, and Enterprise buys a custom annual pool.
Additional watchouts
Credits scale with how many changes and alerts the agent works, so a busy estate can outgrow a plan. Enterprise SSO, the audit log and scheduled agents start at Pro, and region choice at Enterprise.
Overage / add-ons
Not published
Sales call required
Mixed (some tiers require a call)
Free / trial
Evaluation program with 500 credits and no time limit
Lowest paid plan
Team at $700 per month with 1,400 credits
Commercial notes
Independent, with a $4.3 million seed from Zetta Venture Partners. Cleric cites five minutes to root cause, 92 percent actionable findings and more than 200,000 investigations.
Key ambiguities
Plan prices and credit costs are published; what happens past the credit allowance and the Enterprise price are not.
Cancellation / refund
Annual contracts renew automatically after 12 months, with renewal notice and true up terms set in the Order Form.
Support SLA / resale
Team has email support, Pro priority support and Scale a shared Slack channel with Cleric engineers. Enterprise adds a named customer success manager.
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 Cleric
The closest documented capability profiles to Cleric 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 / 14Fuller documented coverage on Human Oversight & Guardrails
- NeuBird11.0 / 14Fuller documented coverage on Human Oversight & Guardrails and Model Flexibility & RoutingCleric vs NeuBird →
- Bluebricks10.5 / 14Fuller documented coverage on Workflow Orchestration and Knowledge Grounding & RAG
- Corelayer11.5 / 14Fuller documented coverage on Workflow Orchestration and Knowledge Grounding & RAG
- Doctor Droid11.5 / 14Fuller documented coverage on Workflow Orchestration and Knowledge Grounding & RAG
- Kubiya11.5 / 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