Back to vendors
S

Sublime Security

Also known as: Sublime Security, Inc., Sublime Autonomous Security Analyst, Sublime Autonomous Detection Engineer

Visit site
Entry priceQuote only for Enterprise agentsFull pricing detail

Sublime Security sells email security for Microsoft 365 and Google Workspace, with AI agents that investigate messages and improve detection coverage. The Autonomous Security Analyst can triage user reports, explain its verdict and apply configured mailbox actions. The Autonomous Detection Engineer can write a rule for a missed threat, test it against the customer's historical mail and route it for review. Enterprise customers can use Sublime Cloud or supported deployments in their own AWS environment.

Sublime Security monitors and protects email for teams using Microsoft 365, Google Workspace and compatible mail systems, and connects through the mail provider's API without MX record changes. Its Autonomous Security Analyst, or ASA, investigates reports sent to the abuse mailbox and messages the platform flags for additional review.

It selects among analysis tools, can run focused investigations in parallel and delivers a verdict, typically in about a minute. User reported messages receive a fuller report with the sender context and evidence the agent examined. The service can make decisions available to a security team through the dashboard, events and API, and sends enriched threat data to SIEM and SOAR platforms through webhooks and S3 exports.

The team can leave ASA in passive mode to produce reports, or switch to active mode and assign actions for particular verdicts. Those actions include quarantine, moving a message to spam, adding a warning banner or clearing a reviewed message from the analyst queue. Uncertain verdicts go to a person. A finding that existing coverage missed can pass to the Autonomous Detection Engineer, or ADÉ. It writes a rule in Message Query Language, tests it against historical mail and returns coverage and results for review. An organization can let qualifying rules activate automatically once they pass its minimum precision threshold.

Enterprise includes both agents and is sold by quote through channel partners. The free Sublime Cloud starting tier covers the first 100 mailboxes. Email data loss prevention, with built in PII, PHI and PCI detections and Microsoft Purview label enforcement, and phishing simulations built from attacks caught in the customer's own mail are Enterprise add ons.

The service can run as managed Sublime Cloud, as a single tenant deployment or self hosted, and supported Enterprise AWS deployments let the customer operate the agents in its own environment and AWS region. ASA also supports AWS GovCloud. Sublime Security holds a SOC 2 Type II report across all five trust criteria, a HIPAA assessment report and an ISO 27001 attestation. Enterprise deployment options include dedicated hosting, single sign on and multifactor authentication. The agents use extracted message context for model inference, and model providers do not retain that input.

The platform fits email security teams that need to reduce manual review and adapt detections using messages in their own environment. Sublime raised a $150 million Series C led by Georgian in October 2025, when it reported four times customer growth that year and no enterprise customer churn since its founding. Elastic used Sublime to automatically remediate more than 2,000 phishing emails before they reached users in its first month.

Snyk pairs ASA for triage with ADÉ for new detection coverage and reclaimed hundreds of analyst hours within weeks of deploying Sublime. Snowflake uses the platform to protect more than 10,000 mailboxes, and Coronis Health cut false positives by 90%. Analysts can retain control of what changes in live email while still letting ASA investigate new reports and ADÉ perform the first review of a new rule. Enterprise teams can purchase email data loss prevention alongside inbound protection, with approval before releasing a message held for review.

Vendor details

Canonical URL

https://sublime.security/

Category

Security / SOC agent

Subcategory

Email threat triage, remediation and detection engineering

Funding status

Venture backed. Raised $150 million in Series C financing led by Georgian.

Company status

independent

Use cases & customers

Primary use cases

Triage user reports and flagged mailApply configured containment or warning actionsGenerate and test rules after missed threats

Target customers

Email security teamsSecurity operations centersDetection engineering teams

Deployment options

Managed Sublime CloudDedicated cloud optionsSupported customer hosted AWS including ASA on GovCloud

Integrations

Microsoft 365, Google Workspace and compatible IMAP sources. REST APIs, Slack and email notifications, webhooks, S3 exports and integrations for security analytics and response.

In practice

A security analyst compares the agent's passive verdicts with their own review before allowing mailbox actions.

A missed message goes to ADÉ, which returns a candidate rule and mailbox test results for an analyst to review.

An IT team enables the Enterprise agents in its own AWS environment and reads the results through its security workflow.

Agentic Index coverage score

12.0 / 14 capabilities · 86%

Integrations & Tool Calling Partial

Active ASA can apply the mailbox action selected for a verdict, including quarantine or moving a message to spam. Sublime connects to Microsoft 365 and Google Workspace and can distribute verdict alerts by webhook, Slack or email. A security team can use those results in its existing response platform. The agent's direct changes are centered on the connected email environment.

Sourcedocs.sublime.security/docs/asa-autonomous-security-analystread 2026-11-20

Workflow Orchestration Full

ASA chooses analysis tools for each message as it investigates and can start focused subinvestigations in parallel. If its verdict confirms an attack existing coverage missed, the analyst agent can send the message to ADÉ. ADÉ then takes over coverage work, testing proposed detection logic and handing results to the security team in a separate review.

Sourcedocs.sublime.security/docs/asa-autonomous-security-analystread 2026-11-20

Knowledge Grounding & RAG Full

ADÉ creates detection coverage using the customer's active mailboxes as a source. Hunts retrieve matching messages across historical email, with a default window of 14 days for a backtest. The agent also uses message dispositions to account for how matches were classified in its environment. ASA uses a separate knowledge base maintained by Sublime's detection team for its investigation reasoning.

Sourcedocs.sublime.security/docs/ade-autonomous-detection-engineerread 2026-11-20

Human Oversight & Guardrails Full

Passive mode records ASA's verdict without changing mail. Active mode applies controls selected by verdict, with uncertain messages routed to a person. ADÉ waits for an analyst's decision by default, and staff can approve or reject a proposed rule, leave comments or cancel a job. An organization can separately enable automatic activation for detections that qualify under its policy.

Sourcedocs.sublime.security/docs/ade-autonomous-detection-engineerread 2026-11-20

Security, Identity & Governance Full

Sublime Security holds a SOC 2 Type II report covering all five trust criteria, a HIPAA assessment report and an ISO 27001 attestation letter from Schellman, with GDPR and Data Privacy Framework participation listed in its trust center. The enterprise plan integrates single sign on and multifactor authentication. Organizations can enable the agents individually and use their existing identity systems for access across the workspace. Access to the agent features is controlled by workspace settings, with ASA inactive before it is enabled.

Sourcesublime.security/plans-detailread 2026-11-20

Observability & Auditability Full

An ASA report can show its verdict, supporting signals and sender context, with the full reasoning available for user reported messages. ADÉ preserves a reasoning summary and the results of the final Hunt. Analysts can follow what the agent reviewed across its rule iterations and see the proposed change with that history, rather than having to infer why a recommendation appeared.

Sourcedocs.sublime.security/docs/ade-autonomous-detection-engineerread 2026-11-20

Memory & State Persistence Partial

Enterprise graymail controls keep an individual's folder preferences over time. They can learn where people want categories of bulk or promotional messages to go, then apply those preferences on later mail in Microsoft 365 or Google Workspace. The preferences are associated with the person, while the graymail controls have no separate expiration setting for that learned state.

Sourcesublime.security/plans-detailread 2026-11-20

Deployment & Data Residency Full

ASA runs in Sublime Cloud and supported Enterprise environments in the customer's own AWS account, including GovCloud. ADÉ also runs in Sublime Cloud and supported customer hosted AWS environments. Where regional inference support is available, ADÉ keeps inference within the US or EU geography. The workload can run in the selected deployment while model processing follows regional availability.

Sourcedocs.sublime.security/docs/asa-autonomous-security-analystread 2026-11-20

Prebuilt Agents / Templates / Packs Full

The Enterprise plan includes Autonomous Security Analyst for mailbox triage and Autonomous Detection Engineer for coverage gaps. The two agents have their own operating views and settings, and a team can enable them according to the work it wants assigned. Sublime also includes standard policies for responding to familiar message outcomes, and teams can adapt those policies to their own mail rules.

Sourcesublime.security/plansread 2026-11-20

Triggers & Channel Coverage Full

ASA can start automatically on abuse mailbox reports or on flagged messages that need additional context following Attack Score. ADÉ can receive confirmed misses either from an analyst's classification or from ASA, and an analyst can also start a review directly in the message view. The configured automations pick up new mail events without a person starting each investigation.

Sourcedocs.sublime.security/docs/asa-autonomous-security-analystread 2026-11-20

Model Flexibility & Routing Partial

ASA and ADÉ use AWS Bedrock for inference. ASA's security language model works from the extracted message context passed by the application. Sublime manages the model connection even when the customer runs the software in its own AWS environment. A security team can set investigation and response policies without choosing a model or providing its own model subscription.

Sourcesublime.security/blog/how-sublimes-ai-agents-are-secure-by-designread 2026-11-20

APIs / SDKs / MCP Extensibility Full

Sublime provides a REST API for security workflows. API clients can retrieve the ASA report for a user reported message and read verdicts in event data. The service also supports direct message ingestion through the API, so a security application can pass mail to Sublime and use the findings in its own work. Organizations can receive notifications without polling the agent's dashboard.

Sourcedocs.sublime.security/reference/retrieveasareportread 2026-11-20

Testing, Debugging & Optimization Full

ADÉ runs a proposed detection rule against historical messages and compares the results with their existing classifications. It refines the rule using the matches and scores its precision before requesting approval. The organization can allow automatic activation only past its configured score, and the final Hunt remains available for review. The standard automatic activation threshold is 99%.

Sourcedocs.sublime.security/docs/ade-autonomous-detection-engineerread 2026-11-20

Browser / Computer-use Partial

The email security platform can render links with a headless browser during analysis. ASA can use information from the linked page when investigating a suspected impersonation. Evidence from the linked page can support the email verdict. An analyst receives the findings in the message report, with no general browser or desktop control service for unrelated security work.

Sourcesublime.security/plans-detailread 2026-11-20

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

Pricing

Quote only for Enterprise agents

Mailboxes, Enterprise modules and deployment

Free tier

What is public

Sublime Security offers a free starting tier in Sublime Cloud for the first 100 mailboxes, aimed at individual security practitioners, with basic trash and move to spam actions. Enterprise is quoted and includes ASA and ADÉ. Email data loss prevention and phishing simulation and training are Enterprise add ons.

Billing mechanics

Enterprise software is sold through channel partners. Customer hosted AWS deployments can incur additional ADÉ processing fees per report.

Sales call required

Yes, required for paid access

Free / trial

Free Sublime Cloud tier for up to 100 mailboxes

Lowest paid plan

Enterprise

Commercial notes

Enterprise support includes 24 hour coverage for Severity 1 issues. ASA is also offered with Enterprise standalone Triage.

Missing data

The quote determines the rate for Enterprise protection, add ons and any charge for detection engineering reports.

Agentic Index verified 2026-11-20

Alternatives to Sublime Security

Other products that do the work Sublime Security does, and what each one is.

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.