AI-generated SDTM, ADaM & TFL specs
Cadence reads your protocol, CRF, and source data to propose mappings — with the rationale behind each.
Agentic Operating System for Clinical Trial Data
From raw clinical data to analysis-ready deliverables
Cadence is an AI-first platform that turns your study data into analysis-ready SDTM, ADaM, and TFL deliverables - with a human in the loop at every step.
Built by a team with experience at Gilead Sciences, Genentech, GRAIL, Delfi, and Medpace
SDTM, ADaM, and TFL work is still slow, manual, and expert-bound
Mapping raw clinical data to SDTM, deriving ADaM, and producing TFLs locks study timelines behind a handful of scarce experts. Cadence changes the economics — whether you're a biotech or a CRO.
From weeks to days
Reach your first analysis-ready mappings and datasets in days, not weeks. 70% reduction in time, while maintaining accuracy.
Experts, multiplied
Your CDISC experts review and approve instead of hand-writing every spec - each one covers more studies. Cadence also functions as an agentic QC programmer.
Audit-friendly by design
Traceable artifacts and audit trails are built into the core design — for specs, code, outputs and agentic actions.
One platform, the full clinical data pipeline
Cadence is AI-first across the stack — from raw source data to CDISC-aligned SDTM, ADaM, TFLs and other deliverables.
Explore data assets using plain languageCreate SDTM mappings in hoursDerive ADaM datasets from SDTM, SAP and TFL shellsGenerate and validate SDTM / ADaM outputsVisualize data traceabilityRun data conformance checks
Built for modern, data-driven teams
AI-generated SDTM specifications· 1 / 5
Context-driven code generation & execution
Package version control, built-in tests, and LLM self-assessment loops. R and Python supported. SAS coming soon.
Traceability and QC built-in
An agentic QC programmer works alongside you in real time, tracing every value back to source.
Explore data in plain language
Ask questions of your study data in natural language — no SQL, no programming, no waiting.
The best way to evaluate Cadence is to watch it work
Book a demoData security & compliance
Cadence is built for regulated clinical data from day one — your data stays yours, and every action is accountable.
Visit Trust CenterNo model training on your data
Your study data is never used to train foundation or third-party models. It is used only to produce your deliverables.
Role-based access controls
Fine-grained roles govern who can view, edit, and approve each study, dataset, and deliverable.
Audit trails for key actions
Every spec change, approval, and generation run is logged — who, what, and when — to support inspection readiness.
Flexible deployment
Deploy in your private VPC to keep clinical data inside your own cloud boundary.
About TrialPath
Clinical trials generate the data that brings new medicines to patients — yet turning that data into regulator-ready deliverables still depends on slow, manual, expert-bound work. TrialPath is building the agentic operating system for clinical trial data: AI that does the heavy lifting, with the humans who own the science always in control.
Founding Team
Harit Nandani
Co-founder & CEO
15+ years in clinical trials data across pharma, diagnostics, and medical device. Previously at GRAIL, Gilead Sciences and Medpace.
LinkedIn →
Tony Wu
Co-founder & Chief Product Officer
10+ years in clinical software and data engineering. Previously at Delfi Diagnostics, GRAIL, and Genentech.
LinkedIn →The TrialPath team also includes two experienced, founding engineers with experience building software at scale in regulated healthcare environments.
Frequently asked questions
Does Cadence support CDISC standards?
Cadence is built to align with current CDISC standards (SDTM, ADaM, CDASH, Define.xml, Controlled terminologies) and produces deliverables structured to those specifications. CDISC is our primary focus today, but the platform is a general agentic system for clinical trial data and is not limited to CDISC.
How do human review and traceability work?
Cadence is agentic but not autonomous. AI drafts every spec, script, and output; humans review, revise, and approve, so nothing is finalized without explicit user approval. Every mapping, derivation, and output stays traceable from a final value back to the raw source column it came from, with an audit trail of approvals.
How are Cadence generated outputs QC'd?
Cadence also includes agentic QC capabilities that help users validate their specs and code - just like a QC programmer would.
Is my data used to train AI models?
No. Your study data is never used to train foundation models, and it is never shared for training purposes. It is used only to produce your deliverables.
How is Cadence deployed and secured?
Cadence supports deployment in your private VPC so clinical data stays inside your own cloud boundary, with role-based access controls and audit trails for key actions. SOC 2 and HIPAA compliance programs are in progress. We'll walk through your organization's specific security and compliance requirements during a demo.
What data formats does Cadence support?
Cadence ingests common clinical data formats including SAS (.sas7bdat) and CSV, and produces standards-aligned outputs (xpt, csv) ready for downstream QC and submission workflows.
How do I get started?
Book a demo and we'll walk you through Cadence on a representative study. We're working with early partners as we finish our first release.
See Cadence on your study
Book a demo. We'll show you SDTM, ADaM, and TFL generation end-to-end — on data that looks like yours.
Loading scheduler…