CISO Tribune

Vendor comparison · 2 companies

Abnormal AI vs Knostic

Abnormal AI sells a behavioral security platform that protects email, identity, and AI. Knostic discovers and secures agents, coding assistants, and their supply chain, from MCP to skills and Code to Cowork. CISO Tribune files both under AI model and agent security. This page sets the two records side by side: what each sells, where it is based, who owns and funds it, the customers it names and its recent news, each fact numbered to its source.

Change the companies

Up to four companies side by side.

  • Abnormal AI
  • Knostic
Abnormal AI vs Knostic: Products, Customers and Funding Compared
Abnormal AIAbnormal AI sells a behavioral security platform that protects email, identity, and AI.abnormal.aiKnosticKnostic discovers and secures agents, coding assistants, and their supply chain, from MCP to skills and Code to Cowork.knostic.ai
What it sells
  • Account Takeover Protection[3][12][19]Detects and remediates compromised Microsoft 365 and Google Workspace accounts.
  • Identity Threat Protection[3][11][12][15]Correlates signals to contain threats before and after authentication.
  • Email Productivity[12][18]Moves low-priority promotional email out of the inbox.
  • Kirin[4][5][6][7][11]Runs on the endpoint inside supported agents. Denies or blocks unsafe agent actions.
  • AgentMesh[4][5][6][7][11]Rates every skill, server, and extension.
Categories
Headquarters
San Francisco, United States[1]Reston, United States[3]
Founded
2018[1]2023[2][7]
Ownership and funding
Privately held[2][12][14]Series D (USD 250 million round), August 6, 2024[2]Privately held[5]
Named customers
  • Domino’s[4]
  • Avery Dennison[5]
  • Valvoline[6]
  • Ingersoll Rand[7]
  • Save the Children[8]
Not stated on its pages
Recent news
  • November 4, 2026Abnormal AI appointed Don Jeter as Chief Marketing Officer, reporting to the CEO and overseeing the global marketing team.[23]
  • September 24, 2026Abnormal AI previewed four AI Security suite products, including AI Agent Security and AI Cloud Security, with general availability planned for Q4 2026.[9]
  • August 27, 2026Abnormal AI expanded its email security platform with Control Center, Email DLP Rules and upgrades to its AI Phishing Coach.[10]
  • August 19, 2026Abnormal AI and OpenAI announced a strategic partnership, with Abnormal expanding its internal use of OpenAI and joining the OpenAI Daybreak Cyber Partner Program as an early security partner.[25]
  • July 29, 2026Abnormal AI extended its behavioral security to cover identities, AI systems and insider threats.[13]
  • June 12, 2026Knostic was recognized in the Cloud Security Alliance's first Agentic AI Security Innovator Market Map.[10]
G2 and Gartner Peer Insights
  • G2: listed, seen October 23, 2026[14]
  • Gartner Peer Insights: Not stated on its pages
  • G2: Not stated on its pages
  • Gartner Peer Insights: Not stated on its pages
Who it fits
Not written yetFrom its own pages, the company sells mainly to developers and teams that use AI coding agents and IDEs, with plans written for individual developers, growing teams, and engineering managers. Its pages describe developers and business users who install agents, extensions, MCP servers, and skills from open ecosystems. It lists an inventory export described as a register of AI systems for compliance and risk reporting. Its stated support covers macOS and Windows.
Last verified
October 10, 2026September 27, 2026

Where they overlap

  • Abnormal AI and Knostic are both filed under AI model and agent security.
  • Both are headquartered in the United States.
  • Both are privately held.

Where they differ, field by field

  • Abnormal AI's primary category is Email security; Knostic's is AI model and agent security.
  • Abnormal AI is also filed under Identity threat detection and response and Workforce AI usage security. Knostic is also filed under AI governance.
  • Abnormal AI lists Account Takeover Protection, Identity Threat Protection and Email Productivity. Knostic lists Kirin and AgentMesh.
  • Abnormal AI is headquartered in San Francisco, United States. Knostic is headquartered in Reston, United States.
  • Abnormal AI was founded in 2018. Knostic was founded in 2023.
  • Abnormal AI's latest recorded funding is Series D (USD 250 million round), dated August 6, 2024.
  • Abnormal AI publicly names Domino’s, Avery Dennison, Valvoline, Ingersoll Rand and Save the Children.

What it sells: Abnormal AI

What Account Takeover Protection does[3][12][19]
Account Takeover Protection learns normal sign-in, device, and behavioral patterns for each user. It remediates a confirmed compromise by signing out all open sessions, blocking access, and forcing a password reset. Abnormal AI says it correlates Microsoft 365 and Google Workspace sign-in signals with communication behavior. It identifies session and token-based attacks that defeat MFA.
What Identity Threat Protection does[3][11][12][15]
Identity Threat Protection correlates signals to contain threats before and after authentication. Abnormal AI says it correlates email, IdP, and SaaS signals into one behavioral model. The model catches privilege misuse and socially-engineered MFA resets.
What Email Productivity does[12][18]
Email Productivity uses behavioral AI to learn what matters to each employee and move low-priority promotional email out of the inbox. It moves detected graymail to a Promotions folder in Outlook or a Graymail label in Gmail. It deploys through API integration with no rules, policies, or quarantine configuration to manage. It supports Passive, Partial, and Active deployment modes.

What it sells: Knostic

What Kirin does[4][5][6][7][11]
Kirin runs on the endpoint inside supported agents and sees the tools, extensions, servers, and skills actually installed and used. Kirin denies a destructive command before it executes, tells the developer why, and logs a Prevented alert for the SOC. Kirin blocks the connection to an MCP server before it is established, after AgentMesh scans and rates the server. Kirin deploys as an extension for Cursor and VS Code.
What AgentMesh does[4][5][6][7][11]
AgentMesh rates every skill, server, and extension. AgentMesh scans MCP servers for prompt injection, malicious behaviour, and supply-chain risk. AgentMesh detects prompt injection in skills, MCP servers, and extensions before install. Every paid AgentMesh plan includes Kirin IDE security.

Questions about Abnormal AI and Knostic

What do Abnormal AI and Knostic sell?
Abnormal AI sells Account Takeover Protection, Identity Threat Protection and Email Productivity. Knostic sells Kirin and AgentMesh.
Are Abnormal AI and Knostic in the same category?
They share AI model and agent security. Abnormal AI's primary category is Email security and Knostic's is AI model and agent security.
Where are Abnormal AI and Knostic headquartered?
Abnormal AI is headquartered in San Francisco, United States. Knostic is headquartered in Reston, United States.
When were Abnormal AI and Knostic founded?
Abnormal AI was founded in 2018. Knostic was founded in 2023.
Who owns Abnormal AI and Knostic?
Abnormal AI is privately held. Abnormal AI's latest recorded funding is Series D (USD 250 million round), dated August 6, 2024. Knostic is privately held.
Which customers do Abnormal AI and Knostic name?
Abnormal AI publicly names Domino’s, Avery Dennison, Valvoline, Ingersoll Rand and Save the Children.
Who are Abnormal AI and Knostic sold to?
From its own pages, the company sells mainly to developers and teams that use AI coding agents and IDEs, with plans written for individual developers, growing teams, and engineering managers.
What have Abnormal AI and Knostic announced recently?
November 4, 2026: Abnormal AI appointed Don Jeter as Chief Marketing Officer, reporting to the CEO and overseeing the global marketing team. June 12, 2026: Knostic was recognized in the Cloud Security Alliance's first Agentic AI Security Innovator Market Map.

About Abnormal AI

Abnormal AI sells a behavioral security platform that protects email, identity, and AI. Abnormal says AI-generated attacks are often too subtle for humans to see and too new for traditional threat intel to detect. Abnormal AI says its platform stops the attacks that slip past legacy gateways, including business email compromise, advanced phishing, vendor fraud, and account takeover. Abnormal AI states its mission is to stop crime with AI.[15]

Abnormal AI says its platform connects AI tools, people, agents, and cloud activity through behavioral context to govern adoption, enforce policies, detect threats, and support investigations. The platform has native API integrations across Microsoft 365, Google Workspace, Okta, CrowdStrike, and the customer's SIEM and SOAR. Abnormal AI says each supported connection helps Abnormal understand activity, investigate threats, and apply the controls the customer's team authorizes. Abnormal AI describes Attune as the behavioral AI model that learns what is normal and detects what does not fit. AI Data Analyst lets users ask plain-English questions and get security insights instantly. AI Data Analyst consolidates insights from across the Abnormal environment into a single source of truth.[15][16][20]

For email, Abnormal AI describes Inbound Email Security as stopping BEC, phishing, and credential theft before delivery. AI Security Mailbox is described as an AI co-worker for email security operations. AI Phishing Coach is described as just-in-time training that adapts to each user's real risk. Messaging Security is described as detecting and remediating malicious content in messaging apps, and Misdirected Email as preventing accidental data leakage from misaddressed email. Email DLP Rules combines a custom policy engine with an AI Triage Agent. Teams can launch Email DLP Rules in passive mode, logging matches without taking action, then move to active enforcement. Email Productivity uses behavioral AI to move low-priority promotional email out of the inbox, and Email Productivity deploys through API integration.[15][17][18][21]

For identity, Account Takeover Protection detects and remediates compromised Microsoft 365 and Google Workspace accounts by learning normal sign-in, device, and behavioral patterns for each user. Account Takeover Protection locks attackers out by revoking active sessions and forcing credential resets. Abnormal AI says Account Takeover Protection connects via API to Microsoft 365 and Google Workspace. Abnormal AI says brute force, credential stuffing, and token or session theft produce sign-ins that satisfy authentication. Abnormal AI says its platform correlates email, IdP, and SaaS signals into one behavioral model, catching privilege misuse and socially-engineered MFA resets. Identity Threat Protection correlates signals to contain threats before and after authentication. Infiltration Prevention is described as detecting synthetic personas and infiltration campaigns before access is provisioned. Posture Management is described as catching risky misconfigurations across the email platform.[3][15][19]

AI Cloud Security is described as detecting AI-driven attacks against cloud identities and workloads and containing threats within the customer's response policies. AI Cloud Security is at the Priority First stage, available for purchase and production deployment to selected customers. AI Employee Guardrails is described as keeping employee AI use on task, protecting sensitive data, and enforcing policies based on the person, tool, and request. AI Agent Security is described as inventorying production agents and verifying that each one behaves as intended. AI Agent Security is at the Design Partner stage, where selected customers evaluate the product and give regular feedback to the product team. AI Employee Guardrails is at the Design Partner stage. AI Security Workbench is described as running AI-powered investigations across petabytes of behavioral events.[3][15][20][21][22]

Abnormal AI says it powers 5,000 customers, including 30% of the Fortune 500.[16][17][18][19][20][21][22]

Abnormal AI on the record

Sources

Abnormal AI full record

  1. [1]Launching 24m series a funding from greylockAbnormal AI · abnormal.ai · published November 19, 2019 · Tier A
  2. [2]Series d 5b valuationAbnormal AI · abnormal.ai · published August 6, 2024 · Tier A
  3. [3]PlatformAbnormal AI · abnormal.ai · undated page, accessed September 27, 2026 · Tier A
  4. [4]Dominos rolls out stronger communications securityAbnormal AI · abnormal.ai · undated page, accessed October 10, 2026 · Tier A
  5. [5]Avery dennison case studyAbnormal AI · abnormal.ai · undated page, accessed October 10, 2026 · Tier A
  6. [6]ValvolineAbnormal AI · abnormal.ai · undated page, accessed October 10, 2026 · Tier A
  7. [7]Ingersoll randAbnormal AI · abnormal.ai · undated page, accessed October 10, 2026 · Tier A
  8. [8]Save the childrenAbnormal AI · abnormal.ai · undated page, accessed October 10, 2026 · Tier A
  9. [9]Abnormal ai expands its ai security suiteAbnormal AI · abnormal.ai · published September 24, 2026 · Tier A
  10. [10]Abnormal expands email security platformAbnormal AI · abnormal.ai · published August 27, 2026 · Tier A
  11. [11]Abnormal extends behavioral security platformAbnormal AI · abnormal.ai · published July 28, 2026 · Tier A
  12. [12]Behavioral AI for Email, Identity, and AI Security | Abnormal AIAbnormal AI · undated page, accessed October 23, 2026 · Tier A
  13. [13]Abnormal AI extends behavioral security to identities, AI systems, and insider threatsHelp Net Security · published July 29, 2026 · Tier B
  14. [14]Abnormal AI Reviews 2026: Details, Pricing, & Features - G2G2 · undated page, accessed October 23, 2026 · Tier C
  15. [15]The Behavioral Security Platform | Abnormal AIAbnormal AI · undated page, accessed November 5, 2026 · Tier A
  16. [16]AI Data Analyst | Automated Security Reporting | Abnormal AIAbnormal AI · undated page, accessed November 5, 2026 · Tier A
  17. [17]Email DLP Rules | Abnormal AIAbnormal AI · undated page, accessed November 5, 2026 · Tier A
  18. [18]Graymail Filtering | AI Email Productivity | Abnormal AIAbnormal AI · undated page, accessed November 5, 2026 · Tier A
  19. [19]Stop Account Takeover | ATO Detection | Abnormal AIAbnormal AI · undated page, accessed November 5, 2026 · Tier A
  20. [20]AI Cloud Security | Abnormal AIAbnormal AI · undated page, accessed November 5, 2026 · Tier A
  21. [21]AI Agent Security | Abnormal AIAbnormal AI · undated page, accessed November 5, 2026 · Tier A
  22. [22]AI Employee Guardrails | Abnormal AIAbnormal AI · undated page, accessed November 5, 2026 · Tier A
  23. [23]Abnormal AI Names Don Jeter CMO to Accelerate Its Expansion in AI SecurityAbnormal AI · published November 4, 2026 · Tier A
  24. [24]Abnormal AI Expands Leadership Team to Advance Behavioral AI PlatformAbnormal AI · published April 29, 2026 · Tier A
  25. [25]Abnormal AI and OpenAI Partner to Accelerate Secure Enterprise AI Adoption and Strengthen Cyber DefenseBusiness Wire · published August 19, 2026 · Tier A

Knostic full record

  1. [1]Knostic - The control plane for agentic AI on the endpointKnostic · undated page, accessed September 27, 2026 · Tier A
  2. [2]Ending llm oversharing we raised 11 million to secure enterprise aiKnostic · knostic.ai · undated page, accessed September 27, 2026 · Tier A
  3. [3]Rsac 2025 innovation sandbox knostic reshaping the access control paradigm for enterprise ai securitySecurity Boulevard · securityboulevard.com · published April 25, 2025 · Tier B
  4. [4]Shadow AI Discovery & Governance - KnosticKnostic · undated page, accessed October 26, 2026 · Tier A
  5. [5]Pricing - KnosticKnostic · undated page, accessed October 26, 2026 · Tier A
  6. [6]AI Agent Security - KnosticKnostic · undated page, accessed October 26, 2026 · Tier A
  7. [7]MCP Security Posture Management - KnosticKnostic · undated page, accessed October 26, 2026 · Tier A
  8. [8]Terms of Service - KnosticKnostic · undated page, accessed October 26, 2026 · Tier A
  9. [9]Knostic Knostic Trust Center | Powered by SafeBaseKnostic · undated page, accessed October 26, 2026 · Tier A
  10. [10]Knostic Featured Across Three Categories in CSA's Agentic AI Security Innovator Market MapKnostic · published June 12, 2026 · Tier A
  11. [11]Prompt Injection Defense - KnosticKnostic · undated page, accessed November 10, 2026 · Tier A

Spot a factual error? Request a correction.