Quan Doan, Vice President of Technology Solutions at SDC, recently led a Live Lounge Presentation at the ACDM 2026 in Berlin, Germany.
In this session, he looked beyond traditional AI capabilities and dashboards available today – with SDC Insights™ 2.0 with SDC Sidekick™ AI- to explore how agentic AI can guide more proactive, confident clinical trial oversight.
✅Watch the full video or simply read the summary below.
What Is Agentic AI?
As AI adoption accelerates in the world around us and in clinical research specifically, the terminology can become noisy. In the context of this talk, agentic AI refers to autonomous software systems designed to perceive their environment, process complex information, and take action(s) toward defined goals—without constant human direction.
A familiar analogy is an automated customer service bot or an AI personal assistant. These systems operate independently, making informed decisions on behalf of the user. In clinical research, agentic AI applies this same principle to data oversight, decision support, and operational execution.
Paradigm Shift – How AI Is Evolving in Clinical Research
Clinical trial oversight has traditionally relied on dashboards that tell us what already happened. Conventional analytics and AI models look backward—learning from historical data to predict future outcomes.
The paradigm is shifting. Today’s expectation is not just prediction, but real-time awareness paired with forward-looking guidance. Rather than waiting for retrospective signals, teams need systems that identify what matters now and what will require attention next.
SDC Insights 2.0 with SDC Sidekick AI – The AI-Native Platform That Exists Today
SDC Insights 2.0 with SDC Sidekick AI (Sidekick) is a clinically trained, AI-native platform designed to surface what is happening now and what needs attention tomorrow.
Sidekick goes beyond static visualization by layering advanced AI capabilities directly onto validated data visualizations. From the start, security and compliance have been foundational—supporting 21 CFR Part 11 and ICH E6(R3) requirements, with transparency and human-in-the-loop validation built in.
The platform operates in a closed-loop, walled-garden environment, ensuring sponsor data remains proprietary and protected. Instead of pushing sensitive data into public AI tools, SDC embeds advanced AI capabilities within a secure environment, minimizing breach risk while preserving performance and flexibility.
This approach allows Sidekick to evolve rapidly alongside advancements in AI—without sacrificing compliance, traceability, or trust.
Advanced Data Cleaning & Automated Queries with Sidekick
Data management is the engine of every clinical trial. While historically treated as a downstream function, the rise of big data and real-time access has elevated data management to a central role in trial execution.
Sidekick functions as a 24/7 digital data manager, continuously scanning trial data for inconsistencies that traditional edit checks may miss. This includes identifying physiological outliers across forms, uncovering hidden patterns, and surfacing potential systemic errors or signals of data fraud.
Importantly, Sidekick removes human bias from the initial detection process. It surfaces potential issues for expert review, enabling data managers to apply judgment where it matters most.
The platform also drafts queries automatically. With a simple verification step, queries can be issued directly—reducing manual effort and cutting query management time by up to 60%.
Data Summaries & Trend Reporting with Sidekick
Sidekick allows teams to ask multi-part questions and receive immediate, contextualized summaries—without waiting for ad hoc listings or custom reports to be programmed.
“In effect, Sidekick acts like having a data manager, biostatistician, and clinical programmer right by your side. Responses can be delivered as narrative summaries, tables, or visual formats, depending on need.”
For teams still building trust in AI, Sidekick offers an added layer of confidence: responses can be flagged for statistical validation by SDC subject matter experts, reinforcing reliability while accelerating insight generation.
AI Studio - Validation of AI Outputs by SDC Experts
As AI becomes integral to clinical trial analytics, model transparency and monitoring are essential.
The built-in AI Studio within SDC Insights 2.0 provides users with appropriate permissions full visibility into model operations, supported by detailed trace logging. Every interaction is documented, including input parameters, processing logic, and outputs.
The platform also tracks model performance over time to detect model drift, ensuring accuracy is maintained as data patterns, languages, and AI models evolve.
Any AI-generated output can be flagged for formal validation by SDC’s statistical programming team, reinforcing trust while maintaining rigorous analytical standards.
Generative AI to Visualize Clinical Trial Oversight
Advances in large language models are happening rapidly. Rather than being constrained by slow validation cycles, Sidekick is designed to adopt innovation safely and efficiently.
Generative UI enables users to describe analytics needs in plain English. Sidekick then generates charts, graphs, filters, and layouts—removing the need for weeks of custom programming.
“This approach enables custom analytics without custom code. Users can refine outputs, save them as reusable components, and pin them to live dashboards that update automatically as new EDC, lab, or external data flows in.”
Over time, one-off questions evolve into permanent, real-time oversight tools—built and maintained through conversation.
Customizable AI Agents – Where is Sidekick Going Next with Agents?
Today, Sidekick is a powerful prompt-based assistant. The next evolution is a specialized suite of customizable AI agents—a digital workforce designed to handle routine clinical data tasks autonomously.
These agents are not static tools. They are purpose-built to pursue outcomes, such as cleaning datasets or flagging protocol deviations, and can be personalized to reflect individual data management styles.
Agents work continuously, reduce operational bias, and complement human expertise—allowing teams to focus on higher-value decision-making.
Agentic AI Data Management Capabilities – What Types of Agents to Create?
Agentic AI enables complex workflows to be broken down into manageable, parallel components—each handled by a dedicated agent. In the context of clinical trial data management and trial oversight, a few types of Agents come to mind:
1) Clinical Data Management (CDM) Agents For Data Mapping
A dedicated agent to manage external data mapping across sources such as central labs, safety systems, and imaging providers. Additional agents can perform reconciliation against EDC data, issue prospective queries, and autonomously map data to SDTM standards.
Together, these agents support end-to-end, continuously operating clinical trial data collection, cleaning, analysis, and standardization.
2) CDM Agents for ConMeds and AE Coding/Mapping
Reconciling concomitant medications, medical history, and adverse events is time-intensive and highly interpretive.
Agentic AI can auto-code indications, standardize terminology, align date ranges, and flag discrepancies across datasets—reducing hours of manual listing review to minutes.
3) CDM Agents to Augment Small or Non-Existent Biostatistics & Programming Teams
For smaller sponsors or biometrics teams without direct access to SAS or R, Sidekick can generate the data joins and program custom listings through conversational prompts. Outputs can be refined, stored, and reused—accelerating insight generation without specialized programming resources.
4) Safety Agents
A safety-focused agent can pull data directly from EDC systems and track safety narratives automatically, reducing manual effort for PVG and safety teams while improving cross-functional visibility.
5) Risk-Based Monitoring Agents
Rather than relying on arbitrary monitoring thresholds, an RBM agent can analyze historical studies and publicly available datasets to establish indication-specific benchmarks. As new study data emerges, thresholds adjust dynamically—supporting more accurate and defensible risk-based monitoring.
The Hybrid Future = AI + Human-In-The-Loop Oversight
Agentic AI delivers speed and scalability in clinical trials, but it must be paired with governance.
Explainable and traceable AI remains central. Models must provide clear reasoning, be monitored for drift, and remain subject to human oversight. As with any powerful tool, responsibility matters.
SDC’s vision is a hybrid workforce—where AI agents perform intensive processing and pattern detection, while human experts validate outcomes, apply judgment, and guide final decisions.
Delivering AI Scalability Today and Tomorrow with Sidekick
SDC Insights 2.0 with SDC Sidekick AI is not just designed for today’s needs. It is built as a platform that evolves—adapting to new AI models, enabling generative interfaces, and supporting customizable agents that accelerate delivery of life-changing therapies.
We partner closely with our clients to build what comes next—combining innovation, compliance, and clinical expertise to empower better clinical decisions.
If you would like to learn more or see Sidekick in action, we welcome continued conversation and collaboration.
Ready To Learn More?
| Speak to the SDC Team Today | Download the Presentation | Book a Demo of SDC Insights with Sidekick |
|
|
|