Graph AI, a startup that uses artificial intelligence to handle patient safety paperwork for drug companies, has raised $13.3 million in a Series A round led by Insight Partners. Existing investor Bessemer Venture Partners also participated. The round comes nearly a year after the company’s seed round in October 2025.
From three hours to ten minutes
The work Graph AI automates is the unglamorous side of medicine. When a patient reports a side effect, pharma companies are legally required to record it, investigate it and file it with regulators, a process known as pharmacovigilance. It is slow, paperwork heavy and expensive.
Graph AI’s platform, Graph Safety, handles those adverse event reports. According to the company, its live deployments have cut case processing time from more than three hours to under 10 minutes, a reduction of over 90%. It also claims operating costs have come down by up to 66%.
Read More: Pixxel Raises $100 Mn Series C Led By Temasek, Seraphim Space
Two modules live, a third this month
The platform runs on two modules today. Intake captures and triages adverse event reports, while Nucleus works as an intelligent safety database that automates case processing. A third module called Report, which handles automated aggregate reporting, is scheduled to launch this month.
Read More: Cradlewise Raises $12 Mn Series A Led By 3one4 Capital, Prudent Investment
The company says its product pairs AI with deterministic controls, validation layers and end to end audit trails. That detail matters in this industry, because regulators need to trace exactly how a safety decision was made, and a black box answer will not survive an inspection.
US and Europe next
Founded in 2024, Graph AI has onboarded pharma and biotech customers across North America and other markets. The fresh capital will fund expansion across the US and Europe and speed up product development. The company is also building Signal, a tool designed to spot emerging safety patterns across cases.
Why it matters
Drug safety is one of those areas where the human cost of slowness is real. A dangerous side effect that surfaces late ends up in more patients. Automation here is not about replacing the medical reviewer, it is about clearing the queue so the reviewer sees the serious case first. If cutting three hours down to ten minutes holds up at scale, that is a rare example of AI doing work that genuinely helps people rather than just cutting a company’s headcount.