InsightFinder Secures $15 Million to Untangle AI's Production Failures

For companies deploying AI, a model's failure is often just the symptom. The real ailment could be hiding in the data pipeline, a server cache, or the complex interplay between new AI agents and legacy systems. InsightFinder, a startup applying 15 years of academic research to this thorny problem, just raised $15 million in a Series B round led by Yu Galaxy to help businesses find the actual cause.

Founded by North Carolina State University professor and former IBM and Google engineer Helen Gu, InsightFinder has monitored IT infrastructure with machine learning since 2016. Its new focus is the reliability of live AI systems. The company's argument is straightforward: you cannot fix what you don't fully see. "To diagnose AI model problems, you must monitor and analyze the data, the model, and the infrastructure together," Gu explained. "It's not always a model problem. Sometimes, it's simply your infrastructure."

She illustrated this with an example: a major credit card client saw its fraud detection model drifting. InsightFinder's platform, monitoring the entire stack, pinpointed the issue not to the algorithm, but to outdated cache in specific server nodes. This holistic view is what Gu calls a necessary "end-to-end feedback loop" from development through to production, a gap many tools miss.

The company's Autonomous Reliability Insights product uses unsupervised learning, predictive AI, and causal inference to analyze data streams and identify root causes. While giants from Datadog to Dynatrace are adding AI features, Gu believes InsightFinder's deep experience with large enterprise environments—clients include UBS, Lenovo, and Dell—provides an edge. "A lot of data scientists understand AI, but they don't understand the system. And a lot of SRE developers understand the system, but not the AI," she noted.

With revenue growing over threefold in the past year, the new funding will support initial sales and marketing hires for the sub-30-person team. The round brings InsightFinder's total raised to $35 million, capital it will use to convince more businesses that fixing AI requires looking beyond the model itself.

Source: TechCrunch