Join our Newsletter — 33% off our NHI Course

Understanding Torq

Torq builds an AI SOC platform for security operations teams that are buried in alerts. AI agents triage, investigate and respond to incidents across the security stack, either fully on their own or with an analyst overseeing each step, so people spend their time on the cases that need judgement.

The Torq AI SOC Platform is aimed at teams that have more alerts than analysts and want the routine part of the job done before a person looks at a case.

At A Glance

VendorTorq
CategorySecurity operations (AI SOC platform)
Identities coveredIdentity response, joiner and leaver, and Just-in-Time access workflows across identity providers
DeploymentSaaS hosted on GCP and AWS
Agentic AIAuto Triage, Torq HyperAgents, Socrates
CustomersBlackstone, Carvana, Chipotle, Marriott, PepsiCo
CertificationsSOC 2 Type II, BSI C5
Market classificationSecurity Operations

What Torq Does

A modern SOC receives more alerts than analysts can read, and alert fatigue means real threats slip through. Torq hands most of that work to AI agents that act in the tools the team already owns. Each agent decision is grounded in what the customer’s environment actually looks like, so an alert about a privileged account is weighed differently from one about a test machine. Teams choose how much autonomy to give, and analysts move from doing the work to directing it.

Products

Auto Triage

Auto Triage is the first filter for a noisy alert queue. It removes duplicate events, screens out false positives and gives each alert an AI verdict that learns from earlier decisions. Analysts can see why a verdict was reached in the audit log and override it by hand.

Case Management

Case Management is the shared record that follows an incident from investigation to remediation. Each case gathers evidence, a timeline and a summary, and cases can be created, assigned and tracked without manual effort.

Torq HyperAgents

HyperAgents take routine SecOps tasks off analysts’ plates. These customisable agents gather evidence, build timelines and recommend actions, and their reasoning stays visible so teams can trust what they do.

Socrates

Socrates coordinates the specialised agents through investigation, evidence gathering and summary. It runs cases end to end and takes instructions in natural language, giving analysts one point of control over the agent workforce.

Hyperautomation

Hyperautomation is the workflow layer that turns decisions into actions across the security stack, built from 300 pre-built integrations and more than 4,000 pre-built steps. It suits teams that want to automate response without writing glue code.

Context Graph And Memory

The Context Graph keeps a transparent, continuously updated model of the customer’s environment, covering identities, roles, privileges and data sensitivity, which agents use to judge what an alert means.

How It Works

Torq is a SaaS platform running on infrastructure from providers such as GCP and AWS, with production environments isolated on the network.

ComponentWhat it does
IngestionTelemetry from across the security stack is ingested and normalised, then correlated and deduplicated before triage.
Data handlingCustomer data is identified, segregated and encrypted at each stage of the data funnel.
AccessUsers sign in through enterprise single sign-on and MFA providers and work under role-based access control. Torq’s own access to production follows Zero Trust and least privilege, and all activity is audited.

Identities And Systems Covered

Identity workflowsResponse to unauthorised access, employee onboarding and offboarding, and Just-in-Time (JIT) access.
Identity providersOkta, Microsoft Entra, Microsoft Active Directory, PingIdentity, OneLogin and JumpCloud.
Identity security toolsSilverfort, Britive, Astrix Security and 1Password.
Cloud platformsAWS, Google Cloud and Microsoft Azure.

Agentic AI Capabilities

Agentic Runbooks And Building

For threat hunting, agentic runbooks search historical cases and summarise patterns. The Agentic Builder lets AI agents help engineers build new integrations and workflows.

Autonomy Levels

Teams decide how far the agents may go, and Torq reports that over 90% of cases are remediated autonomously. Response can run completely autonomously, or with human-on-the-loop oversight where an analyst watches and can step in, and the audit log with manual override in Auto Triage keeps every agent verdict reviewable.

Integrations

AreaIntegrations
EDR and XDRCrowdStrike, SentinelOne, Microsoft Defender, Cortex XDR, Carbon Black
SIEMSplunk, Microsoft Sentinel, IBM qRadar, Elastic Security, Exabeam
Cloud securityPrisma Cloud, Orca Security
TicketingAtlassian Jira, ServiceNow, PagerDuty
CommunicationsSlack, Microsoft Teams, Google Chat

Company Background

Founded2020.
HeadquartersAmericas headquarters in Denver, Colorado, United States.
TeamMore than 430 employees across the United States, Canada, Israel and Europe.
LeadershipOfer Smadari (co-founder and CEO) and Josh Morris (Chief Revenue Officer).
FundingA reported $140 million Series D in January 2026, led by Merlin Ventures at a $1.2 billion valuation, for $332 million raised in total. Other investors are Greenfield, Evolution Equity Partners, Notable Capital, Bessemer Venture Partners and Insight Partners.
AcquisitionJit, a Boston-based company, in May 2026.
RecognitionRecognised as an AI SOC platform by Gartner, SACR and KuppingerCole in 2026.

Note: This overview is compiled from Torq’s publicly available information as of 3 November 2026. It is provided for information only. NHI Mgmt Group does not endorse any vendor, and nothing on this page represents the opinions of NHI Mgmt Group.

Explore More

Browse vendors

Free weekly newsletter

Subscribe to the NHI & AI Identity Journal

The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.

Bonus 33% off our NHI Course when you subscribe.

Book a demo

See the platform

Tell us a little about you and we’ll pass your request to the vendor’s team.

Fill in all fields marked * and tick the consent box to continue.