Mimecast
Also known as: mimecast.com, Mimecast Agent Risk Center, Mimecast AI Security, Mimecast Mihra, Mimecast Incydr, Mimecast Apex
Mimecast sells email, collaboration, data protection and human risk security under the Mimecast Apex platform, and its AI Security line governs how employees and their agents use AI. Agent Risk Center inventories AI desktop apps and MCP connections into GitHub, Salesforce and Slack, ties each agent to the person behind it, and enforces an AI Rulebook that sanctions, blocks or nudges. The Mihra Investigation Agent triages Incydr data protection alerts and recommends next steps, and an Incydr MCP server lets teams query that data from their own AI tools.
Mimecast sells security for the human side of the enterprise under its Apex platform: email and collaboration security against phishing, impersonation and account takeover, data protection through Incydr, security behavior management with real time nudges and training, data governance with archiving and eDiscovery, and AI security. More than 42,000 organizations use Mimecast through its regional affiliates and resellers, and Incydr, which arrived with the 2024 acquisition of Code42, protects source code and intellectual property at Snowflake and Banked. Mimecast cites more than 350 integrations.
The AI Security line watches how people and the agents they deploy use AI. Agent Risk Center, which runs on the Incydr endpoint agent and browser extension already in place, builds an inventory of AI desktop apps, exposes MCP connections that link agents into systems such as GitHub, Salesforce and Slack, captures copy, paste and upload activity into generative AI tools, and maps each item to the employee driving it, with a departmental heatmap of unsanctioned use. Each agent's actions are weighed against the behavioral profile of the person who deployed it.
The AI Rulebook marks tools and connections as sanctioned, unsanctioned or pending for the whole company or a department, group or individual, and either blocks a tool or nudges the user toward an approved one. On the analyst side, the Mihra Investigation Agent triages and prioritizes Incydr alerts, enriches events with context, summarizes them and recommends next steps and controls with its reasoning shown, while analysts decide what happens. The Incydr MCP server, hosted in the Incydr cloud and reached with a client ID and secret, lets analysts query alerts and events in plain language from Claude, ChatGPT or Copilot, and Governance, Compliance and Insights archives Claude Enterprise conversations for legal hold and eDiscovery.
Every plan is quoted. Data Protection comes in Core with 50 AI credits a month, Pro with an endpoint agent, 100 credits and Agent Risk Center, and Max with 150 credits in a FedRAMP authorized environment. Mimecast holds a SOC 2 Type II report and lists ISO 27018, HIPAA and CSA STAR in its trust center, and runs regional infrastructure in the UK, Germany, Canada, Australia, South Africa and Singapore.
Incydr access runs on roles such as Insider Risk Admin, Analyst and Read Only, with SAML single sign on through Okta, Entra ID, Google, Ping or OneLogin and SCIM provisioning that maps groups to roles. Agent Risk Center opened in beta in July 2026, and Mihra runs on a model Mimecast built in house.
It fits organizations already on Mimecast email and Incydr that want to see and police the AI tools and agents their staff adopt, with insider risk alerts triaged in the same place. It fits less well for teams that want agents taking response actions on their own, those choosing their own model inside the product, or buyers without the Incydr endpoint agent.
Vendor details
Canonical URL
https://www.mimecast.com
Category
Security / SOC agent
Subcategory
Email, data and human risk security suite with discovery and policy control of employees' AI tools and agents, plus an AI investigation agent for insider risk alerts
Funding status
Private company, owned by Permira since 2022, with its US base in Lexington, Massachusetts. Mimecast operates through regional affiliates including Mimecast South Africa (Pty) Limited, Mimecast Australia Pty Ltd., Mimecast Germany GmbH and Mimecast Canada Limited.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Agent Risk Center runs on the Incydr endpoint agent and browser extension and maps MCP connections into systems such as GitHub, Salesforce and Slack. Mimecast cites more than 350 integrations with API and MCP connections, a public API that SOAR and XDR tools call, and an Incydr MCP server for Claude, ChatGPT and Copilot. Governance, Compliance and Insights ingests Claude Enterprise through Anthropic's Compliance API.
In practice
A software company turns on Agent Risk Center and finds 40 developers running desktop AI apps, six of them with MCP connections into the production GitHub organization. The security team marks two tools unsanctioned in the AI Rulebook, which blocks them, and nudges the rest toward the approved assistant.
A sales director uploads a pipeline export into an unsanctioned chatbot the week after giving notice. Incydr raises the alert, and the Mihra Investigation Agent summarizes the file movement, the departure context and the risk, and recommends revoking the tool and opening a case for the analyst.
A legal team uses Governance, Compliance and Insights to place three executives' Claude Enterprise conversations under legal hold and search them alongside their email for an eDiscovery request.
Sources & related URLs
Related / legacy domains
Research sources
Agentic Index coverage score
8.0 / 14 capabilities · 57%
| Integrations & Tool Calling | Full |
|---|---|
|
Agent Risk Center reaches into the systems employees' agents touch, exposing MCP connections into GitHub, Salesforce, Slack and similar tools, and enforces its decisions on the endpoint and in the browser through the Incydr agent and extension already installed. Mimecast cites more than 350 integrations with API and MCP connections, and SOAR and XDR tools call its API to pull message parts or decode links during response. The wider Incydr program links email, endpoint, identity, SIEM and SOAR signals. Sourcemimecast.com/blog/mimecast-extends-data-protection-to-the-agentic-enterpriseread 2026-11-18 |
|
| Workflow Orchestration | Partial |
|
AI Security runs a fixed sequence over the AI tools and agents it finds: discover what is running, govern what is sanctioned, detect what is risky and respond to what matters, with the AI Rulebook applying the response. The Mihra Investigation Agent works through triage, enrichment and summary for each alert. Customers do not build their own flows across several agents, and the Configuration and Detection agents that would extend the chain are still planned. Sourcemimecast.com/products/ai-securityread 2026-11-18 |
|
| Knowledge Grounding & RAG | Partial |
|
Decisions draw on the inventory of AI apps and MCP connections, each employee's behavioral profile, Incydr file events and the ApexIQ intelligence layer shared across Mimecast products. Through the MCP server analysts can also upload policies such as an acceptable use policy and check events against them. There is no knowledge base of the customer's own documents that the agents search and cite as a retrieval store. Sourcemimecast.com/blog/mimecast-extends-data-protection-to-the-agentic-enterpriseread 2026-11-18 |
|
| Human Oversight & Guardrails | Partial |
|
The AI Rulebook is a set of guardrails on employee AI use: each tool or connection is sanctioned, unsanctioned or pending, and the response is either a block or a nudge toward an approved tool for lower risk cases, applied company wide or to a department, group or person. The Mihra Investigation Agent works alongside analysts and only recommends, so people make every response decision. There is no queue where a proposed action waits for a person to approve it. Sourcemimecast.com/blog/mimecast-extends-data-protection-to-the-agentic-enterpriseread 2026-11-18 |
|
| Security, Identity & Governance | Full |
|
Mimecast holds a SOC 2 Type II attestation covering security, availability, processing integrity and confidentiality, and its trust center also lists ISO 27018, HIPAA and CSA STAR, while Data Protection Max runs in a FedRAMP authorized environment. The Incydr console, where Agent Risk Center and Mihra's findings live, assigns access through roles such as Insider Risk Admin, Insider Risk Analyst, Insider Risk Read Only, Insider Risk Respond and Audit Log Viewer. Users sign in over SAML with Okta, Entra ID, Google, Ping or OneLogin and are provisioned through SCIM, with identity provider groups mapped to Incydr roles. Sourcemimecastsupport.zendesk.com/hc/en-us/articles/42665963712275read 2026-11-18 |
|
| Observability & Auditability | Full |
|
Agent Risk Center keeps a record of what each AI tool and agent does: the desktop apps in use, the MCP connections into business systems, the copy, paste and upload activity into generative AI tools and the employee behind each one, with a heatmap of where unsanctioned use clusters by department. Agents' actions are scored against the deploying person's behavioral profile. Mihra's recommendations show their reasoning so analysts can audit them. Sourcemimecast.com/blog/mimecast-extends-data-protection-to-the-agentic-enterpriseread 2026-11-18 |
|
| Memory & State Persistence | Not documented |
|
Incydr keeps the event history, alerts and cases that later investigations read, and the inventory keeps the record of tools and connections over time. That is the security record itself. The Mihra Investigation Agent has no memory of its own, with a scope and lifetime, that carries analyst preferences or earlier decisions into its next triage. Each triage starts fresh. Sourcemimecast.com/blog/mimecasts-spring-launchread 2026-11-18 |
|
| Deployment & Data Residency | Full |
|
Customers are hosted on regional infrastructure built around local residency expectations: dedicated regions in the UK and Germany, independent deployments in Canada, Australia and South Africa, and a full platform on AWS in Singapore for customers in Southeast Asia and Hong Kong. Data Protection Max runs in a FedRAMP authorized environment for US public sector buyers, and Agent Risk Center works through the endpoint agent and browser extension on employees' own machines. Sourcemimecast.com/blog/mimecast-launches-singapore-hosting-for-southeast-asia-and-hong-kongread 2026-11-18 |
|
| Prebuilt Agents / Templates / Packs | Partial |
|
One ready agent ships today, the Mihra Investigation Agent, with Configuration and Detection agents announced for later quarters. The AI Rulebook starts every discovered tool as sanctioned, unsanctioned or pending for the security team to settle, and policies are assembled per department, group or person. There is no catalog of prebuilt policies, playbooks or agents to adopt for other jobs. Sourcemimecast.com/blog/mimecasts-spring-launchread 2026-11-18 |
|
| Triggers & Channel Coverage | Full |
|
Once enabled, the inventory populates from the endpoint agent and browser extension and keeps updating as employees install AI apps, add MCP connections or paste and upload data into generative AI tools. Rulebook decisions fire on that activity as it happens, without an analyst starting them. Incydr alerts arrive at the Mihra Investigation Agent, which triages and prioritizes them automatically before an analyst opens the queue. Sourcemimecast.com/blog/mimecast-extends-data-protection-to-the-agentic-enterpriseread 2026-11-18 |
|
| Model Flexibility & Routing | Not documented |
|
The Mihra Investigation Agent runs on a dedicated large language model that Mimecast built in house and tailored to Incydr's data, and customers do not pick a different model for it. Analysts who prefer Claude, ChatGPT or Copilot reach Incydr data through the MCP server from inside those assistants instead, which changes where they ask questions rather than which model powers Mimecast's agent. Sourcemimecast.com/blog/mimecasts-spring-launchread 2026-11-18 |
|
| APIs / SDKs / MCP Extensibility | Full |
|
Mimecast runs a public API that SOAR, XDR and custom scripts call, for example to fetch message parts or decode rewritten links. The Incydr MCP server is hosted in the Incydr cloud, set up with a client ID, secret and base URL, and connected through a desktop extension so Claude Desktop and other MCP clients can query alerts and file events in plain language. It is included in every Incydr SKU at no extra cost. Sourcemimecastsupport.zendesk.com/hc/en-us/articles/46837183313299read 2026-11-18 |
|
| Testing, Debugging & Optimization | Not documented |
|
Neither the customer's AI agents nor Mimecast's own Mihra agent can be put through test cases, red teaming or scored evaluations in the product. The AI Risk Assessment offered on the AI Security page is a sales led review of a company's AI exposure, and Mihra's quality shows up only in how analysts judge its recommendations case by case. No replay or comparison of agent versions exists. Sourcemimecast.com/products/ai-securityread 2026-11-18 |
|
| Browser / Computer-use | Not documented |
|
The browser extension watches and controls what employees do with AI tools, and email browser isolation renders risky links away from the user's device, but no Mimecast agent drives a browser or desktop to click through another application. Mihra works on Incydr data inside the console rather than by operating other systems. Any web step in an investigation stays with the analyst. Sourcemimecast.com/blog/mimecast-extends-data-protection-to-the-agentic-enterpriseread 2026-11-18 |
|
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; Core, Pro and Max plans per solution
Subscriptions to Core, Pro or Max tiers for each solution, with monthly AI credits included in Data Protection
Included quota
Data Protection Core includes 50 AI credits a month, Pro 100 and Max 150.
What is public
Mimecast prices every plan on request. Each solution comes in Core, Pro and Max tiers. Data Protection Pro adds an endpoint agent, 100 AI credits a month and Agent Risk Center, and Max runs in a FedRAMP authorized environment with 150 credits.
Billing mechanics
Solutions are bought separately or together: Email and Collaboration Security, Data Protection, Security Behavior Management and Data Governance each have their own tiers, and AI Security is sold through a demo. The Incydr MCP server comes with every Incydr SKU at no extra cost.
Cost watchouts
Agent Risk Center comes with Data Protection Pro and relies on the Incydr endpoint agent, and Data Protection Core requires Email and Collaboration Security.
Variable cost rationale
Plans are quoted subscriptions with a fixed monthly allowance of AI credits.
Additional watchouts
AI credits are counted per month, and what one credit buys is not stated.
Sales call required
Yes, required for paid access
Free / trial
No free plan; a 30 day proof of value for Agent Risk Center on request
Lowest paid plan
None; every plan is priced on request
Commercial notes
Mimecast offers Incydr customers a 30 day proof of value of Agent Risk Center.
Missing data
Mimecast gives no prices for any plan.
Related vendors
- 7AI — Agentic SOC from the Cybereason founders: domain specialized agents…
- Abnormal AI — Behavioral AI email security with three named autonomous agents: a…
- AirMDR — AI native MDR whose AI analyst investigates every alert and writes…
- Airrived — Agentic OS that unifies SOC, GRC, IAM, vulnerability management, IT,…
- Andesite — Human-AI SOC platform where customers build their own agents and…
- Anvilogic — Agentic security operations platform that works on top of existing…
Alternatives to Mimecast
Other products that do the work Mimecast does, and what each one is.
- IdiraIdentity security platform extended to AI agents
- Horizon3.aiNodeZero autonomous penetration testing across internal
- OktaIdentity platform extended to AI agents
- Operant AIRuntime security for AI agents, MCP servers and employee AI tools
- Qevlar AI
- HiddenLayerAI security platform for the customer's own models and agents