TrueFoundry + Straiker: AI Gateway Security from Build to Production

Enterprises are moving AI agents into production, and thesecurity reality looks very different from a prototype. Agents don't justanswer questions. They call models, use tools, and act on data across manyteams and frameworks. Traditional gateway controls answer only part of thesecurity question: authentication, rate limits, and access policies decidewhether an agent may act, but not whether an authorized agent has beenmanipulated. The harder question is this: how do you bring AI gatewaysecurity to every model call without rebuilding protection inside every agent?

At TrueFoundry, our approach is to keep the execution layeruniform and let teams plug in the security system they choose. That's why we'reannouncing an integration between the TrueFoundryAI Gateway and Straiker, bringing agentic AI security intothe infrastructure enterprises use to build, deploy, and scale AI applicationsand AI agents. Platform teams run the gateway, security teams own the policies,and application teams stop rebuilding security in every agent.

Introducing TrueFoundry AI Gateway

The TrueFoundry AI Gateway establishes a single, governedentry point for all model and agent requests. Applications and agents no longertalk directly to model providers. They talk to the gateway, which gives teamsaccess to 2,500+ LLMs through one OpenAI-compatible API and applies routing,rate limits, budgets, and request tracing centrally. With under 3 ms ofinternal latency, even at enterprise scale, it can sit directly in theproduction path of every request. That makes the gateway the one consistent placeto enforce AI guardrails: built in, from externalsecurity providers, or custom.

For security leaders, this is the point where AI protectionstops being a per-application project and starts behaving like infrastructure.

Introducing Straiker

Straiker adds agent-specific security to that control point.Agent attacks rarely fit inside one prompt and response, so Straiker inspectsfull agentic traces across turns and sessions. This gives its detection layerthe context to evaluate what an agent is doing, which resources it is using,and how its behavior changes over time.

Discover AI maps the agentic ecosystem acrossan organization. For organizations standardizing on TrueFoundry, it adds asecurity view of what has been deployed and what each agent can reach. AscendAI tests agents through automated, multi-turn adversarial campaigns,offering remediation plans and benchmarking against today's most popularcompliance frameworks. Defend AI inspects live agent activity andapplies policy to prompts, responses, tool calls, and multi-step workflows. Itmaps an enterprise's agentic blast radius across identities, surfaces, agents,models, and tools, and monitors and blocks malicious and suspicious activity.Straiker's published benchmark reports a 98.1%true-positive rate and a 0.7% false-positive rate across itsevaluated threat categories. Straiker is designed for low-latency runtimeenforcement, with detection generally completing in less than 300 milliseconds.

Blocking one request may not stop a compromised agent.Straiker's Agentic Kill Switch provides an escalation path fromtransaction-level blocking to agent-level containment: depending on thedeployment, teams can revoke tool access, suspend sessions, freeze memory, ortake an individual agent or agent fleet offline.

Both companies already support shared enterprise customers,including Automation Anywhere and Comcast.

How TrueFoundry and Straiker worktogether for AI gateway security

An AI gateway is a natural enforcement point for productionAI security, yet most teams still approach AI agent security one application at a time.They typically end up choosing between two unsatisfactory options:

-       Build and maintain security separately insideevery agent.

-       Rely on gateway controls that can't tell when anauthorized agent has been manipulated.

On the TrueFoundry side, Straiker plugs into the AI Gatewayas a guardrail that checks prompts on the way in and model responses on the wayout, as well as any toolscalls connected to the agent infrastructure. There is no adapter orwrapper service to build or run. Teams can turn it on for specific models oracross traffic through a gateway policy, with no changes to application code.

On the Straiker side, security teams set the controls foreach agent in the Straiker console. Straiker evaluates prompts and responsesagainst those controls and returns an allow or block verdict, which the gatewayenforces, giving applications runtime AI protection where their traffic alreadyflows.

Getting started with Straiker onTrueFoundry

In Straiker Defend, add an agent and select TrueFoundry fromthe available gateway integrations. Security teams set that agent's controls inthe Straiker console, including which detections to block, and review everyfinding there. Straiker's TrueFoundry integration guidecovers this side of the setup.

In the TrueFoundry AI Gateway, add Straiker as a guardrailon the models or traffic you want to protect. Start in Audit mode to see whatStraiker flags without blocking anything, then switch to Enforce once theresults look right. Each check also appears in TrueFoundry's request traces, soplatform and security teams review the same activity. The TrueFoundry Straiker guide walks through eachstep.

Conclusion

For AI and security leaders, the TrueFoundry and Straikerintegration makes AI gateway security part of the same layer that alreadyroutes and governs production AI. Platform teams keep one governed gateway,security teams own the policies, and application teams ship agents withoutrebuilding protection in each one. The partnership is intentionally composable:TrueFoundry routes and enforces, Straiker evaluates and decides. Teams canbuild, deploy, and govern with TrueFoundry, then discover, test, defend, andcontain with Straiker.

See Straiker in action or book aTrueFoundry AI Gateway demo.

Frequently asked questions

What is AI gateway security?

AI gateway security means enforcing security controls at theAI gateway, the shared layer every model and agent request passes through, soprompts and responses are checked in one place instead of in every application.With TrueFoundry and Straiker, that layer enforces Straiker's agent-specificverdicts.

Does the Straiker integration needextra infrastructure?

No. The TrueFoundry AI Gateway connects to Straikerdirectly, so there is no adapter or wrapper service to build or run.

Which models does the Straikerintegration work with?

Any LLM the TrueFoundry AI Gateway routes to, from OpenAI,Anthropic, and Google models to self-hosted ones. Guardrails work the same wayfor every model behind the gateway, so a single Straiker guardrail can coverall of them.

Can TrueFoundry run in my own cloudor on-premises?

Yes. TrueFoundry runs in your own VPC, on-premises,air-gapped, or across hybrid and multi-cloud setups.

Does TrueFoundry meet SOC 2, HIPAA,and GDPR standards?

Yes. TrueFoundry is built to meet SOC 2, HIPAA, and GDPRstandards, with SSO, role-based access control, and immutable audit loggingbuilt in.

Related reading

-       Whatis an AI Gateway? The Complete Guide (2026)

-       How to Configure Guardrails on the AI Gateway

-       WhatIs an LLM Gateway? A Complete Guide