InsightFinder Secures $15M to Debug AI Agent Failures

The role of observability tools has evolved once more. While the market for solutions that ensure the reliability of technology systems has expanded over time, the focus has steadily shifted from "tracking everything" to "managing complexity and controlling costs." Meanwhile, the rapid influx and adoption of AI agents within enterprises has introduced an entirely new category of workloads that require observation.

InsightFinder AI, a startup built on 15 years of academic research, is intimately familiar with this challenge.

The company has utilized machine learning since 2016 to monitor, identify, and proactively resolve IT infrastructure issues. It is now addressing contemporary AI model reliability concerns with an AI agent solution capable of handling everything from detection and diagnosis to remediation and prevention.

Founded by CEO Helen Gu, a computer science professor at North Carolina State University and former IBM and Google employee, the company recently secured $15 million in a Series B funding round led by Yu Galaxy, TechCrunch has learned exclusively.

According to Gu, the most significant challenge facing the industry today extends beyond monitoring and diagnosing AI model failures. It involves diagnosing how the entire technology stack functions now that AI is an integral component.

"To effectively diagnose issues with AI models, you must monitor and analyze the data, the model, and the underlying infrastructure collectively," Gu explained to TechCrunch. "The problem isn't always isolated to the model or the data; it's often a combination. Sometimes, the root cause is simply your infrastructure."

Gu illustrated this with a real-world example: One of InsightFinder's clients, a major U.S. credit card company, experienced drift in its fraud detection model. Because InsightFinder was monitoring the company's entire infrastructure, it identified that the model drift was caused by outdated cache data on specific server nodes.

"A common misconception is that AI observability is limited to LLM evaluation during development and testing. In reality, a robust AI observability platform should provide end-to-end feedback loop support covering development, evaluation, and production stages," she stated.

InsightFinder's latest product, named Autonomous Reliability Insights, achieves this by combining unsupervised machine learning, proprietary large and small language models, predictive AI, and causal inference. According to Gu, this foundational layer is data-agnostic, enabling the system to ingest and analyze complete data streams, gather signals, and correlate and cross-validate them to pinpoint root causes.

The observability landscape is now crowded with competitors vying for a share of the new market created by the proliferation of AI tools. After nearly a decade in operation, InsightFinder competes with companies like Grafana Labs, Fiddler, Datadog, Dynatrace, New Relic, and BigPanda, all of which are developing capabilities to address the novel challenges posed by AI.

However, Gu remains confident. She asserts that InsightFinder's expertise, experience, and customizability form a strong competitive advantage. "We rarely lose customers to competitors... It ultimately comes down to the insights. The issue is that many data scientists understand AI but not the system, while many site reliability engineering (SRE) developers understand the system but not the AI... They often miss the intrinsic relationships between them."

Today, InsightFinder's customer portfolio includes UBS, NBCUniversal, Lenovo, Dell, Google Cloud, and Comcast. Gu attributes this success to a decade of effort dedicated to understanding the needs of large enterprise clients.

"It has involved collaborating closely with our Fortune 50 customers to refine our approach and understand the enterprise environment requirements for deploying these models," she said. "We've been working with Dell to deploy our AI systems globally for some of our largest clients. This isn't something you can achieve by simply applying a foundational AI model to machine data."

Gu reported that the company's revenue stream is "strong," having grown "more than threefold" in the past year. In fact, the company was not actively seeking this Series B funding; investors approached them after InsightFinder secured a seven-figure deal with a Fortune 50 company within three months.

InsightFinder will use the new capital to make its first sales and marketing hires to expand its team of fewer than 30 people and invest in its go-to-market strategy. To date, the company has raised a total of $35 million.