Securing the Internet: How Outtake's AI Agents Dismantle Cyber Attacks at Scale with Inngest
“Throttling and durable execution are essential for automating cybersecurity at scale. We have several rate limits in a variety of places, both incoming in terms of APIs and outgoing in terms of token spend. Inngest is helping our developers confidently create and deploy highly complex agentic AI workflows faster than ever.”
Outtake is a cybersecurity platform leveraging always-on AI agents to scan millions of digital surface areas (webpages, app store listings, ads, and more) per minute to detect and dismantle digital impersonations. Its innovative approach drastically reduces manual labor for enterprise security teams, shrinking threat remediation timelines from weeks to minutes and helping counter millions in fraud losses. Through real-time threat classification and automated response, Outtake is transforming digital security and supercharging organizations in protecting their online identities at Internet scale.
The challenge of context engineering: scanning millions of domains and webapps daily.
Human oversight alone is no longer enough to keep pace with today's AI-driven cyber threats. Attackers now spin up convincing clones of enterprise websites in minutes,deploying fake login, checkout, and support pages at scale to trick users and steal their credentials. Outtake counters this with 24/7 protection. By embedding AI agents throughout its system, Outtake continuously hunts, classifies, and remediates attacks without ticket backlogs. A core component of their system's architecture is Inngest, ensuring Outtake's models get the right data, in the right format, at the right moment. That precision enables complex security tasks to be handled autonomously, keeping clients and users safe online.
Outtake leverages a comprehensive multi-step context-building pipeline implemented through Inngest. The primary objective of this system is to recognize and counter harmful impersonations across the Internet.
Outtake systematically scans the digital landscape, spanning search engines, domains, websites, and forums, to uncover unstructured data that may signal threats. This information flows through advanced detection pipelines designed to detect identity attacks. Outtake's AI agents classify abuse types ranging from sophisticated phishing and impersonation to intellectual property violations. Guided by each client's brand guidelines, IP policies, and enforcement preferences, the agents automatically take actions to remediate threats.
Building a system of this complexity, with hundreds of data sources with autonomous AI-focused workflows layered on top means managing massive swings in compute while ensuring rock-solid reliability:
There are so many limits to juggle in such a variety of places. We face them on the incoming side in terms of datasource building and on the agentic side in terms of token usage. Further in any complex workflow, AI agents can behave unpredictably as we configure our systems. If those issues surface without control, they can quickly disrupt progress. Having fine-grained control over errors and how they propagate is incredibly valuable. Function invocations are just as important. Our pipelines are intricate, with many interdependent components and their own concurrency challenges, all chained together.
Outtake overcomes these challenges with Inngest's powerful workflows and Flow Control features:
- Durable workflows help Outtake's data and AI pipelines recover seamlessly from external interruptions, resuming workflows from the last successful checkpoint. This not only conserves resources but also accelerates processing.
- Throttling manages the load of incoming data streams by spreading them out over time, thereby preventing managing scale limits intelligently.
- Batching speeds up Outtake's processing of millions of daily events, grouping similar information to reduce compute and database load.
Inngest: Outtake's Event-Driven Core
Outtake's powerful threat scanning engine leverages Events, yet another key feature of Inngest.
Built on an event-driven architecture, Outtake leverages powerful patterns, including batching, debouncing, and event replay, to ensure that automated security workflows run smoothly. Inngest's highly reliable Event API ensures that every sent event is persisted, serving as the durability layer of Outtake's processing steps:
Our agents reason and act on millions of datapoints at once. With Inngest, we can queue up millions of events without worrying about loss or disruption. That reliability is everything.
With events' guaranteed reliability and availability, Inngest's Replay feature gives Outtake a powerful way to maintain continuity when workflows are interrupted:
Replays and error handling are so critical. Being able to replay functions is incredibly useful because outcomes can be unpredictable. Whether it's an outage or data inconsistency somewhere that affects function runs, having the ability to retry that work ensures we process everything end-to-end without missing a thing.
Inngest's Replay feature enable engineering teams to recover from intermittent failures by replaying a set of events. Replaying events matches the functions' Flow Control, enabling a smooth recovering of operations, even for millions of events.
Building AI Agents with Inngest
Outtake's approach demonstrates that effective AI agents require a robust architecture to handle large datasets, manage rate limits, and ensure reliability. Using Inngest's workflows, throttling, and event-driven design, Outtake developed a cybersecurity system that processes hundreds of thousands of digital attack surfaces daily without missing a single threat.
Ready to build your own AI products with Inngest? Book a call with our experts today to learn how we can help you orchestrate reliable, scalable AI workflows for your business.