Announcing TriClick™ 3.0 — Stronger Generation, Full Submission Documents, and Protocol-to-CRF
Today we’re announcing TriClick 3.0. Highlights across generation quality, submission documentation, reviewer transparency, and a new protocol-to-CRF capability.
Stronger Core Generation Across SDTM, ADaM, and TLF
Each of the three clinical generation agents was benchmarked against internal accuracy suites.
- The SDTM Agent now reaches nearly 79% final-output accuracy.
- The ADaM Agent reaches mid-80% final-output accuracy.
- The TLF Agent shows targeted benchmark improvements from the high-50s to the low-90s on average, with the largest lift exceeding 55 percentage points.
Submission Documents, Not Just Datasets: ADRG, SDRG, and Define.xml
TriClick 3.0 adds support for the Analysis Data Reviewer’s Guide (ADRG), Study Data Reviewer’s Guide (SDRG), and Define.xml — the reviewer-facing submission artifacts that accompany datasets in an FDA filing.
Audit Package as a Core Deliverable
Alongside the datasets and TLFs, TriClick 3.0 delivers a customer-facing Audit Package with:
- Review dashboard
- Evidence manifest
- Audit graph
- Execution trace
- Review packet
- Prioritized worklist
Clinical teams see what was generated, what was checked, and what still needs review.
Lineage Graph — Source to Output
A new Lineage Graph traces how the AI moved from study inputs to final deliverables, capturing source data, specifications, CRFs, shells, intermediate decisions, generated code, checks, and final outputs.
Transparent Reasoning by Default
For every supported output, TriClick 3.0 exposes the source evidence used, intermediate artifacts, assumptions, checks performed, and review status. No black-box answers.
New Capability: Protocol-to-CRF (DM Agent)
TriClick 3.0 adds the Data Management Agent, which turns a study protocol PDF into a structured first-draft CRF. It extracts visits, assessments, SOA, eligibility, safety, dosing, and follow-up requirements.
- Every generated field links back to its source protocol evidence
- Reviewers can leave field-level comments; the DM Agent generates revision proposals with before/after comparisons for human approval
- The agent auto-detects protocol language and supports a separate CRF display language
- Teams can export the CRF document, eCRF draft, and SOA matrix for downstream review
Review-Ready, Not Blind Automation
TriClick 3.0 does not claim every generated result is 100% correct. It generates high-quality outputs while making uncertainty, evidence gaps, and review needs visible. For pharma and CRO teams: faster generation with a stronger foundation for QC, handoff, and audit prep.
Learn More
- Talk with our team about TriClick 3.0 on your sample dataset → hillresearch.ai/contact-us
- Platform overview and technical notes → hillresearch.ai/triclick
- Sign up → app.hillresearch.ai/signup
Congratulations to our engineering, biostatistics, and regulatory teams who shipped this release.