Product

One proven engine drives four regulated documents, and nothing here is sold before it ships.

The deterministic engine is the core of TriloDocs — in production on live studies today. Every document type below states exactly which architectural layer produces it, its validation status, and what is guaranteed.

● IN PRODUCTION — available today ○ EXPANSION SEQUENCE
FLAGSHIP

Clinical Study Report

IN PRODUCTION ENGINE: DETERMINISTIC · LAYER 1 VALIDATION PACK AVAILABLE GAMP 5 CAT

TriloDocs reads your source outputs — TFLs, SAP, protocol, synopsis — and drafts the CSR the way an experienced regulatory writer does: rule by rule, signal by signal, with every numerical claim transcribed from source and traced to its exact table. There is no probabilistic inference anywhere in the data path, which means the output is reproducible, auditable, and validatable.

  • Source documents processed and first draft produced in minutes, not weeks
  • Zero hallucination in data processing — guaranteed by architecture, not model behaviour
  • Identical inputs produce identical outputs — fully reproducible for QC and re-runs
  • Decision criteria are transparent and customisable to your SOPs and style guide
How a CSR is produced

Source documents Your inputs

TFLs · SAP · Protocol

Deterministic engine Layer 1 · owns all data

Rule-based signal identification. Values transcribed, cross-checked, and traced — never generated.

Language layer Layer 2 · phrasing only

Narrative readability. Architecturally prevented from producing numerical or safety-critical content.

Verified draft + audit trail Regulatory-ready

Every sentence traceable to source. Re-verified by Layer 1 before release to your writers.

COMPANION DOCUMENTS

Three further documents, each labelled with exactly what produces it.

Different documents carry different risk profiles, so they use different layers of the platform — and we say which, on the tin. Your AI governance committee should never have to ask.

Lay Summary

CTEG-compliant plain-language summaries
Status● In production
EngineLanguage layer, data-locked
NumbersLocked to source via Layer 1
ReviewHuman-in-the-loop required
Converts trial results into patient-readable summaries covering all 10 CTEG components, GDPR- and HIPAA-conscious by design. The language layer writes the prose; every figure it uses is supplied and verified by the deterministic engine — the LLM cannot introduce a number of its own.

Patient Narrative

Individual narratives for safety reporting
Status● In production
EngineLayer 1 data + language layer
NumbersTranscribed, never generated
ScaleAll patients, one run
Generates consistent, structurally identical narratives across every patient in the trial from verified case data. Because the data layer is deterministic, narrative N of 400 is held to exactly the same standard as the first — the failure mode of manual narrative writing at scale.

Informed Consent Form

Patient consent documentation
Status● In production
EngineLanguage layer, template-bound
StructureRegulatory-aligned templates
ReviewHuman-in-the-loop required
Drafts consent documentation in clear, patient-accessible language with consistent structure across all studies, aligned to regulatory requirements from the first version. Reduces first-draft time so your team's effort goes into study-specific judgement, not boilerplate.
Why we label the engine on every product: not all documents carry the same risk, and pretending otherwise is how AI vendors lose the trust of QA teams. Where the language layer leads, we say so, we lock it away from your data, and we build the human review step into the workflow rather than around it.
EXPANSION SEQUENCE

Where the engine goes next — for the customers already on it.

2026 · Q3

Investigator Brochure

Comprehensive investigator information for trial sites and ethics submissions, built on the same traced data layer.

2026 · Q3

Module 2.7.4 Clinical Safety

Summary of clinical safety, synthesised across the full safety dataset with the same traceability guarantees as the CSR engine.

2026 · Q3

TriloDocs MCP foundry

Summarise sections. Compare runs. Surface related findings across documents.

2027 · Q1

Module 2.7.3 Clinical Efficacy

Summary of clinical efficacy, built on the same deterministic core as the rest of the Module 2.7 set.

2027 · Q2

Module 2.7.2 Clinical Pharmacology

Summary of clinical pharmacology studies.

2027 · Q2

Module 2.7.1 Biopharmaceutic Studies

Summary of biopharmaceutic studies, completing the Module 2.7 Clinical Summary set.

2027

Module 2.5 Clinical Overview

2.5.2 Biopharmaceutics · 2.5.3 Pharmacology · 2.5.4 Efficacy · 2.5.5 Safety.

Existing customers get early access to each release and their validation documentation is extended, not repeated.

TriloDocs
See TriloDocs in Action

Book a demo or speak with our team.

We reply within one working day.