June 2026 • PharmaTimes Magazine • 30-31
// AI //
Time travel
How agentic AI is shifting data management from time-consuming investigation to confident confirmation
AI is reshaping clinical data management (DM), moving the industry away from time-consuming, inefficient processes and towards continuous, intelligent oversight.
Agentic AI can transform one of the most resource-intensive workflows in clinical data management – manual listing review.
Manual listing review is complex, time-consuming and judgement based. Inefficient processes result in a high investigation burden, review bottlenecks, limited scalability and the introduction of greater risk into clinical data management.
Validated agentic AI offers the opportunity to shift DM work from laborious investigation to confident confirmation, improving data quality while accelerating query cycles and timelines.
‘Evidence-backed findings also reduce subjective interpretation and variability, leading to a higher percentage of queries being accepted’
It empowers DM teams to identify true discrepancies; surface supporting evidence and translate findings directly into action.
In this article we explore the limitations of the current manual listing review life cycle, outline the benefits of using agentic AI in manual listing review and discuss the regulatory compliance that must be in place.
Manual listing review remains one of the largest consumers of skilled DM capacity, with low precision and high wasted effort.
This results in lost capacity, slow query cycles and timeline risk, highlighting the need for solutions that provide measurable impact on DM.
Yet traditional solutions are failing to overcome these challenges.
Instead of focusing on depth, EDC-native approaches – which are not designed for cross-domain expertise – prioritise platform breadth and simple edit checks. This leaves the majority of judgement-based reconciliation untouched.
Similarly, classic machine learning approaches are optimised for breadth, dashboards or retrospective insights rather than giving sponsors full control and validated precision.
Outsourcing shifts the effort of manual listing review but fails to tackle the inefficiency, while also reducing sponsor visibility and control over data quality decisions. This limits real-time data visibility and threatens consistent decision logic.
Success in a pressurised DM environment requires strategic thinking about where AI can be applied most effectively.
Agentic AI combines LLMs, machine learning and natural language processing to pursue complex goals with limited supervision.
It is better suited to the complex, judgement-based data checks required for DM than EDC edit checks, broad analytics platforms, CRO-led processes, internal builds or solutions relying on machine learning alone.
In DM, agentic AI can unlock capacity, improve data quality and enable faster execution without changing workflows, ownership or regulatory posture. This removes AI governance friction and protects sponsor control.
Agentic AI should no longer be viewed as an experimental add-on. Instead, sponsors and regulators should recognise it as an increasingly essential element of infrastructure for demonstrating data integrity and accelerating submission readiness – a collaborator rather than a tool.
In contrast to manual review, which results in around 40% precision, agentic AI can identify true discrepancies with 80% or more precision, eliminating time spent on non-issues.
This allows a higher proportion of DM effort to shift to confirmation and high-value analysis, unlocking measurable capacity without increasing staff headcount.
Auto-generated, editable query text enables faster progression from discrepancy identification to query submission and reduces back-and-forth with sites due to clearer, evidence-backed queries.
This results in shorter overall query cycle time with less rework and a focus on execution speed and quality, not just detection.
Evidence-backed findings also reduce subjective interpretation and variability, leading to a higher percentage of queries being accepted on the first pass and fewer rejections, clarifications or rework cycles.
By applying standardised decision logic consistently across studies and teams, agentic AI improves consistency in how judgement-based issues are raised.
Late-stage listing backlogs threaten database lock and lead to last-minute cleaning marathons.
By focusing on the 60–70% of manual effort that is cross-domain and high variance, agentic AI offers a depth-first approach that concentrates effort where delays are most likely to occur.
This results in fewer listing review bottlenecks, earlier identification of high-risk discrepancies that could delay database lock and increased predictability in late-stage data cleaning.
AI adoption fails without regulatory clarity. Any system must be transparent, auditable, defendable and 21 CFR Part 11 compliant.
There must be full traceability from signal to evidence to action to outcome, with documented reasoning for all AI-assisted outputs.
Restricting agentic AI models to read-only access can also ensure no data modification or imputation risk is introduced.
Developing these solutions internally requires multi-year R&D investment and ongoing maintenance, with a high risk of never reaching production-grade precision.
Instead, sponsors should look for immediately usable, validated capability backed by a proven scenario library developed and tested in real studies.
Intelligent query detection that combines agentic AI with precision monitoring and adaptive scenarios reduces wasted effort while maintaining sponsor control and regulatory clarity.
It eliminates manual investigation at scale by automating the most time-consuming parts of clinical data review, improves data quality and accelerates query cycles with reliable, evidence-backed findings that adapt as studies change.
When used correctly, agentic AI delivers automation without risk, combining regulatory-ready auditability with full transparency and control.
Organisations that embrace agentic automation early will see dramatic reductions in manual review burden.
Varun Cruz is Founder of Neuvior