Agentic AI Systems to Reshape Toxicology Workflows Soon
A new presentation at PharmSci 360 will detail how auditable AI models are set to automate chemical safety assessments.
Updated on Oct. 21, 2026 in Artificial Intelligence
Live Poll
Do you trust artificial intelligence systems to make accurate and auditable scientific risk assessments?
Insilica founder Thomas Luechtefeld is set to present on agentic AI systems at the AAPS 2026 PharmSci 360 conference in New Orleans on October 28, 2026. The presentation will highlight how AI-driven workflows are increasingly central to toxicology research and decision-making.
Why it matters
Single predictive models are evolving into broader decision-making ecosystems that must prioritize regulatory compliance and traceability. This shift toward agentic systems, which can audit their own data retrieval processes, is expected to fundamentally alter toxicology team workflows within the next two years.
The RASAR machine learning model achieved 87% balanced accuracy across nine OECD test types while evaluating a dataset of 190,000 chemicals. These results contrast with traditional single-model methods by enabling automated, auditable retrieval from toxicology databases.
The players
Thomas Luechtefeld
Founder and CEO of Insilica who developed the RASAR machine learning framework for chemical evaluation.
Insilica
A developer of AI-driven toxicology tools that maintains a presence at booth 919 at the upcoming PharmSci 360 conference.
The details
Agentic AI systems — software capable of executing complex sequences of tasks autonomously — are being deployed to organize document information and query specific databases. These systems are designed to be auditable, meaning they explicitly document which databases were queried and the success status of each step. This transparency is critical for risk and hazard assessments, as it allows researchers to trace exactly how the system arrived at its conclusions.
Timeline
October 25-28, 2026: AAPS 2026 PharmSci 360 conference takes place.
October 28, 2026: Thomas Luechtefeld presents at the conference in New Orleans.
1 to 2 years: Expected timeline for shifts in toxicology team workflows.
The Tech Race
The transition to agentic AI marks a departure from static predictive models that have historically anchored chemical safety assessments under OECD test guidelines. As toxicology teams shift toward these traceable, autonomous systems, Insilica is positioning its platform to become an industry standard for audit-ready regulatory documentation.
Toxicology researchers can access the company's ToxIndex database at no cost to integrate these findings into their own workflows. Over the next two years, these agentic systems are expected to change how teams manage, verify, and document hazard decisions in regulatory filings.
The takeaway
The field is moving toward systems that prioritize the audit trail as much as the predictive result. Watch for upcoming research and case studies that demonstrate how these agentic tools perform against complex, real-world regulatory dossiers.
Further reading
For more on the latest research in this field, visit the Artificial Intelligence section.
Source note: This article includes information reported by BioPharm International.
Live Poll
Do you trust artificial intelligence systems to make accurate and auditable scientific risk assessments?
