What TriClick Is
TriClick™ is an AI platform for clinical statistical programming. It takes the documents and datasets a study already produces — protocols, SAPs, specifications, CDISC datasets, TLF shells — and runs AI-assisted workflows over them to generate, review, and package statistical deliverables.
How work is organised
Work happens inside a study. Each study holds collections — typed groups of files, so the platform knows an ADaM spec from a set of TLF shells. You run workflows against those collections, and each run publishes its results as a new collection alongside the inputs that produced them.
That structure is the reason everything stays traceable: inputs and outputs sit side by side, nothing is overwritten, and every deliverable can be traced back to the run and the inputs that made it.
What's New in 3.0
Latest release3.0 extends the platform from producing datasets to producing the full reviewer-facing package that accompanies them — and hands the last mile of code refinement back to your programmers.
- Refine in your own IDE. Generated programs open in VS Code, where your team reviews and finalises them. Details
- Submission documents. define.xml, SDRG, and ADRG generated alongside the datasets. Details
- Audit package and output provenance. Every output carries a "how this was produced" view covering stages, inputs, and checks. Details
- Stronger core generation. Accuracy improvements across SDTM, ADaM, and TLF generation.
Full release notes are in the TriClick 3.0 announcement.
What You Give, What You Get
The shortest answer to "what does this actually do":
You provide
- Protocol and Statistical Analysis Plan
- Dataset specifications (SDTM / ADaM)
- Raw or CDISC datasets
- TLF shell documents
TriClick returns
- SDTM and ADaM datasets
- Tables, Listings and Figures
- The R or SAS programs behind them
- define.xml, SDRG and ADRG
- Review findings and an audit trail
Capability
Generation — SDTM, ADaM, TLF
The core of the platform. Generation runs as a chain, and you can enter it at whichever point matches where your study already is.
- SDTM datasets — programs and datasets generated from your SDTM specification.
- ADaM specifications — drafted from your SDTM datasets. Limited release.
- ADaM datasets — programs and datasets from SDTM plus the ADaM spec.
- Tables, Listings and Figures — from ADaM datasets and a TLF shell document.
- SAPs and TLF shells — drafted from a protocol when you don't have them yet. Limited release.
R or SAS, your choice
Generation produces working programs, not just rendered output, and you pick the language per run. The programs download alongside the results, so you can rerun or audit them in your own validated environment.
Runs are observable
Every run appears on the study's Runs page with per-output progress and logs. A failed output can be retried on its own rather than re-running the whole job, and re-running never overwrites earlier results — it creates a new run.
Generation feature
New in 3.0Refine in VS Code — The Last Mile
TriClick generates the study code. The expensive part — reviewing it, correcting it, and signing it off — stays with your statistical programmers, and the VS Code extension lets them do that work in the editor they already use.
- Open generated programs directly. SDTM, ADaM, and TLF programs open in VS Code from the platform.
- Bring your own AI. Point your team's local AI coding assistants at the code, right alongside it.
- Edits sync back. Changes return with full version history, tagged human or agent — nothing leaves the audit trail.
- Regenerate in-platform. Revise derivations, then regenerate outputs from the platform.
Capability
Automated Review
Generated outputs go through an automated cross-check before they reach formal QC. The review covers statistics, formatting, and consistency, and publishes its findings as review reports attached to the output collection.
- Findings are severity-ranked and categorised, so a programmer can triage rather than read everything.
- Each finding points at the specific output it came from.
- Review can be re-run across the project as outputs change.
Review-ready, not blind automation
TriClick does not claim every generated result is correct. It aims to produce high-quality output while making uncertainty, evidence gaps, and review needs visible — so QC effort goes where it is actually needed.
Capability
Submission Documents
Datasets are only part of a filing. TriClick generates the reviewer-facing documents that accompany them, driven from your dataset specification:
- define.xml — for SDTM and ADaM.
- SDRG — Study Data Reviewer's Guide.
- ADRG — Analysis Data Reviewer's Guide.
Reviewer's Guides are generated as editable documents you finish in the in-app editor, then export in submission format.
Capability
Structured Editors
The documents that drive generation are edited in the platform rather than passed around as files, so they stay machine-readable and versioned.
- TLF shells — define each output's number, caption, population, layout, and footnotes, validated as you type.
- Statistical Analysis Plan — a section-by-section editor over the canonical SAP outline, exportable to Word.
- Dataset specifications — edit SDTM and ADaM specs by datasets, variables, codelists, methods, and where-clauses, with CDISC controlled terminology pinned per study.
- Reviewer's Guides — finish generated SDRG and ADRG drafts and export them.
Specification edits are checked on save, so a broken spec cannot be committed silently.
Working in TriClick
Outputs, Versions & Releases
Outputs
Results are previewable in-app — tables, documents, datasets — with the generated programs downloadable next to them.
Versions
Every file is a versioned artifact. Uploading a replacement or re-running a workflow adds a version rather than overwriting; each version records who, when, and which run. Editor saves are versioned the same way.
Releases
A release freezes the state of a study's collections under a name — a data cut, or a submission milestone. You can browse a release read-only exactly as the study looked at that moment, and export its contents for delivery without ongoing work changing them. This is how you answer "what exactly did we deliver?"
Working in TriClick
Access, Roles & Audit
Role-based access
Access is role-based, covering tenant administration, delegated study administration, day-to-day statistician and programmer work, review-oriented QA access, and read-only access. Users are invited by email or arrive through your single sign-on, and can be grouped to manage access at study granularity.
Audit trail
- Run history — a permanent record of every run: inputs used, who started it, status, timing.
- Output provenance — a "how this output was produced" view covering stages, inputs, and checks, with a downloadable review-ready summary.
- Application activity — logins, uploads, runs, and admin changes, with per-user drill-down.
Together, roles, versions, runs, and activity mean every deliverable traces back to who ran what, on which inputs, when.
Reference
Standards & Compliance
- CDISC — SDTM and ADaM, with controlled terminology pinned per study.
- 21 CFR Part 11 — designed to meet electronic records and signatures requirements.
- GCP-ready — audit-defensible, versioned outputs with full provenance.
Reference
Plans & Access
Self-serve individual plans — TriClick Go, Plus, and Pro — let biostatisticians, programmers, and clinical scientists evaluate the platform on real workflows without a procurement cycle. Plans are metered in table credits, with shared projects, audit logs, and governed exports at the higher tiers.
Team, department, CRO, and pharma sponsor rollouts — including custom pricing, volume credits, single sign-on, and dedicated onboarding — are handled directly.