Expert Intelligence raises $4.7M to automate decision-making in laboratory environments – News

Expert Intelligence, a startup building artificial intelligence systems to automate decision-making in regulated lab environments, today announced it has  raised $4.7 million in a seed funding round.

All labs where analytical testing is involved in drug manufacturing, pharmaceutical, food and beverage safety involve limited time and resources. After any instrument processes a batch, some runs are clean, but some are ambiguous.

Given a limited amount of reagents to work with, an analyst must make a judgment about every outcome. Lalin Theverapperuma, Ph.D., co-founder and chief executive of Expert Intelligence, told News that a simple analogy is similar to banking before ATMs: Most verification steps were manual for every verification, because trust and consistency were not automated yet.

Expert Intelligence built a new AI-driven platform with an AI model at the core called the Limited Sample Model, a new approach designed to learn how expert analysts make decisions with a small number of samples. Unlike many chemistry and life sciences models in the industry, which are based on generative AI large language models and machine learning models that require massive datasets, LSM is built for regulated environments where data is scarce.

“We built LSM so regulated labs can scale expertise with accuracy, transparency, and audit readiness from day one,” said Theverapperuma. “Our models learn directly from how experts interpret raw signals, allowing labs to increase throughput without introducing new compliance risk.”

Theverapperuma went on to explain that LSM learns patterns in instrument signals, direct raw data from mass spectrometer readouts, time series data, expert decision policies — for example, how experts weigh evidence and decide — and the audit trail needed for defensible reporting.

The whole paradigm wraps back to the scarcity of data and the critical importance of expert attention.

“Because lab data is sparse, we designed LSM to learn from tens of representative examples, often around 30 for a workflow, then automate the decision layer end to end, from raw instrument output to report-ready decisions in minutes,” Theverapperuma added.

The company began deploying the model in early 2025. Since then Expert Intelligence has secured customers within analytical testing workflows within many regulated lab environments to support use cases including automated result review, anomaly detection and expert-level decision consistency.

The funding, led by Sierra Ventures with participation from TSVC and Acorn Pacific Ventures, will be used to accelerate customer expansion in pharma, deepen independent laboratory systems and expand into additional lab-driven industrial domains over time.

Image: News/Microsoft Designer