Zscaler 블로그

Zscaler 블로그의 최신 업데이트를 이메일로 받아보세요

Products & Solutions

The key of AI: How Agentic Tuning can make your detection strategy sing

image
SAM STRAKA
January 26, 2026 - 4 분 소요
>The key of AI: How Agentic Tuning can make your detection strategy sing

Remove unwanted alerts from your environment using plain language

This article was originally published by Red Canary, which is now part of Zscaler.

Stop me if you’ve heard this one before: security alerts can be noisy. Mostly, these noisy alerts are communicating information that is, on average, important information. But the reality of your specific organization isn’t described by an average. The noisy alert just doesn’t “get” your established controls and accepted risks. For many organizations, alerts in their queue represent an overwhelming number of “authorized, expected in the environment” activity—noise that burns time, drains confidence, and obscures what truly matters.

Staying in tune

At Red Canary, we’ve honed our analytics to filter out an incredible amount of that noise while ensuring we don’t miss threats. If there’s something important buried in all those alerts, we have to catch it. That means Red Canary must cast an extremely wide detection net; we turn the sensitivity up to 11 so that nothing slips past.

It’s kind of like listening to a hundred different songs at once while making sure you don’t miss one, very important, drum beat.

So we apply multiple layers of tuning at different points in the detection and investigation lifecycle—both global tuning that applies to everyone and per-customer tuning that’s specific to your organization. That gets us to our stunning, industry-leading low rate of false positives.

But what about the rare things that resist tuning? The behavior that is normal and expected in your environment but has every hallmark of suspicious activity? Behavior that, in any other organization, would definitely be an indicator of something concerning? To keep you safe, Red Canary has to look into it to keep you covered. Even this small subset of false positives means work for your security team.

Now, with Agentic Tuning, you can filter out highly specific and difficult-to-tune alerts with a few plain-language sentences typed into Red Canary. We’ve created a system combining AI agents, human expertise, and our extensive experience in detection analytics. It’s the last puzzle piece for achieving next-level tuning.

Why this matters

False positives are costly

They consume analyst capacity, slow response, and erode trust in your detections.

Every customer is unique

Risk tolerance, security architecture, access policies, device management, and compensating controls vary widely.

Tuning shouldn’t be a black box

You deserve transparent, auditable logic that aligns with your policies—and you should be able to change it quickly.

The last puzzle piece: Plain language suppression enabled by AI agents

Agentic Tuning is driven by two different systems: Customizations and the Threat Review Agent in Red Canary. Customizations is a new portal experience that lets you provide explicit, auditable guidance Red Canary uses during threat investigations. Those instructions are evaluated by our Threat Review Agent—an AI system that consults your active customization items and the full investigation context to recommend whether a threat should be suppressed or published.

This is agentic AI built for security operations: a purpose-built agent that reasons over telemetry and your explicit guidance, explains its decisions, and acts consistently at scale.

How it works

  • Point us in the right direction with a few sentences, and we’ll take it from there.
  • Every customization item flows through a review process to ensure appropriateness and actionability.
  • All changes are audited so you know who is adding or changing your customizations. Because our number one priority is securing your environment, Red Canary may override customizations to alert you to suspected threats when necessary.

What does that look like? Here’s an example:

 

Screenshot of plain language instructions to an AI agent in the Red Canary Portal What happens next
  • After evaluation by the Threat Review Agent, threats that match the customization criteria are suppressed and don’t alert your team, letting you focus on other work.
  • Suppressed threats are still available for you to audit and review, including an explanation which references the customization item used.

 

What you’ll get

Fewer false positives

You’ll have more time to focus on the real threats. Early access customers have seen up to an 80 percent reduction in false positives in identity-related detections.

Policy-aligned detections

Your expectations drive outcomes. As your environment evolves, your customizations evolve with it.

Get started

  • Visit the new Customizations section in your Red Canary Portal.
  • Add instructions to reflect suppressive policies you want enforced.

Security teams shouldn’t have to choose between complete coverage and a manageable signal. With Customizations and Agentic Tuning, you get both—an AI-driven, agentic layer that adapts to your environment and delivers the right detections at the right time.

form submtited
읽어주셔서 감사합니다

면책 조항: 이 블로그 게시물은 정보 제공만을 목적으로 Zscaler에서 작성하였으며, 정확성, 완전성 또는 신뢰성에 대한 어떠한 보증도 없이 "있는 그대로" 제공됩니다. Zscaler는 제공된 정보를 기반으로 수행된 조치나 오류 또는 누락에 대해 어떠한 책임도 지지 않습니다. 이 블로그 게시물에 링크된 제3자 웹사이트나 리소스는 편의를 위해서만 제공되며, Zscaler는 해당 웹사이트의 콘텐츠나 관행에 대해 책임을 지지 않습니다. 모든 콘텐츠는 예고 없이 변경될 수 있습니다. 이 블로그에 접속함으로써 귀하는 본 약관에 동의하며 귀하의 필요에 따라 정보를 확인하고 사용할 전적인 책임이 귀하에게 있음을 인정합니다.

Zscaler 블로그의 최신 업데이트를 이메일로 받아보세요

양식을 제출함으로써 귀하는 당사의 개인정보 보호정책에 동의하는 것입니다.