Onea AI Technology Platform

Expand Analytical Capacity with AI. Keep Critical Judgment Human.

Clinical development is challenging not simply because of the volume of information, but because critical evidence, data discrepancies, assumptions, and dependencies are often distributed across nonclinical, CMC, PK, clinical, and regulatory sources.

The Onea AI Technology Platform applies AI to evidence extraction, cross-document comparison, evidence integration, development analysis, and scenario evaluation—supporting professional work across development readiness, FIH strategy, clinical study design, and IND preparation. Final scientific, clinical, statistical, and regulatory judgments remain human-led.

AI-Amplified. Human-Led. Evidence-Grounded.

Critical IND-to-FIH Questions Are Often Distributed Across Multiple Sources and Documents.

Four Core Platform Capabilities

01 | Evidence Intelligence

Identify the evidence that truly matters to the development question.

Systematically review Sponsor documents, CMC, nonclinical studies, existing clinical data, scientific literature, and relevant regulatory sources to identify critical data, cross-document inconsistencies, evidence gaps, and unresolved uncertainties.

Source interrogation
Cross-document comparison
Evidence traceability

02 | Development Intelligence

Turn evidence into the questions that determine whether development can move forward

For a defined development milestone, organize the available evidence, unresolved issues, and key dependencies to clarify which issues may affect further development and which decisions require additional evidence or specialist review.

Readiness assessment
Cross-evidence relationship assessment
Development impact analysis

03 | Study Design Intelligence

Make study-design assumptions easier to examine—not decisions for AI to make.

Integrate scientific evidence, relevant regulatory frameworks, statistical assumptions, and operational feasibility to support comparison across design scenarios and help Sponsors understand the trade-offs among population, dose/regimen, endpoints, sample-size assumptions, and other design choices.

Assumption structuring
Scenario modeling
Cross-domain design review

04 | Document Intelligence

Make documents carry established development decisions—not conceal unresolved issues.

Translate confirmed evidence, development rationale, and study-design decisions into clinical and regulatory documents, while using source traceability and cross-document consistency checks to reduce discrepancies in data, assumptions, and rationale across deliverables.

。

Structured authoring
Source traceability
Cross-document consistency

What We Deliberately Do Not Delegate to AI

AI Does Not Independently Determine Whether Evidence Is Sufficient

AI Does Not Independently Make Clinical Study Design Decisions

AI Does Not Replace Medical / Toxicology / Statistics / Regulatory Specialist Judgment

AI Does Not Make Final Development Decisions on Behalf of the Sponsor

AI Output Is Not Evidence.

AI Capabilities Must Be Paired with Controlled Human Review

01｜Source Traceability

Material evidence and claims retain a traceable source basis.

02｜Separation of Candidate Findings from Final Judgment

Candidate issues identified by AI are not automatically accepted as formal conclusions.

03｜Appropriate Expert Confirmation

Issues requiring domain-specific judgment are escalated to the Sponsor, CRO, or appropriate specialist for confirmation.

04｜Explicit Retention of Uncertainty

When available evidence cannot resolve an issue, the uncertainty remains explicit rather than being filled in by AI-generated answers.

Identify Potential Amendment Drivers Earlier

The Onea Clinical AI Platform is designed not only to accelerate Study Design and Protocol development, but also to identify unresolved design assumptions, cross-functional inconsistencies, key dependencies, and execution constraints before the design is finalized.

Across population, dose / regimen, endpoints, statistics, safety, and operational feasibility, the platform helps teams identify potential amendment drivers before Protocol Freeze.

Data Security & Confidentiality

Onea treats the confidentiality and appropriate use of Sponsor data as fundamental requirements of an AI-assisted workflow. Project materials are handled within the agreed scope of work, and the approach to AI-assisted analysis is determined according to data sensitivity, client requirements, and the applicable working environment.

Where confidential, non-public, personal, or otherwise restricted information is involved, appropriate data-handling conditions and usage boundaries are confirmed before the relevant workflow begins.

Validation Framework Designed in Alignment with Core GxP Principles

Onea Clinical’s AI validation workflow is built around traceability, version control, evidence provenance, independent review, human oversight, and controlled qualification records.

We do not validate only whether AI can generate an answer. We validate whether:

the reasoning process is traceable, the evidence is controlled, independent reviewers can assess the work separately, and final conclusions are adjudicated by humans within clearly defined accountability boundaries.

Prospective Validation of AI-Assisted Clinical Development

Onea validates not only whether AI can generate a Study Design or Protocol, but whether the complete chain—from controlled development evidence and design derivation to Sponsor-facing documents—can remain consistent, traceable, and independently reviewable without access to the historical answer.

Prospective outputs are completed and frozen before the historical comparator is opened, reducing the risk of retrospectively adjusting the design after the historical answer is known.

Complete and freeze the prospective outputs first. Open the historical comparator only afterward.

Case 01 | PER-19A Prospective Evidence-to-Protocol Qualification

Validation Objective

Under a predefined evidence cutoff and controlled case-specific evidence set, complete the full derivation chain—from development evidence and Study Design to a Sponsor-facing Synopsis and Full Protocol—before opening the historical comparator, and subject the prospective outputs to independent review.

Validation Results

PER-19A completed a full prospective Evidence-to-Protocol qualification.

With the historical comparator withheld throughout the prospective phase, controlled development evidence successfully progressed through independent analysis, Study Design, Sponsor-facing Synopsis, and Full Protocol development. Independent review confirmed cross-document fidelity, protocol executability, and professional-authority boundaries. The historical comparator was opened only after all prospective outputs had been completed and frozen.

The purpose of historical comparison is not to replicate a known answer. PER-19A recovered multiple high-information features of the historical trial architecture while retaining material differences that could not be uniquely determined from the controlled evidence.

What we validate is not simply whether AI can generate a Protocol, but whether the complete Evidence → Study Design → Synopsis → Full Protocol derivation chain can remain consistent, traceable, and independently reviewable before the historical answer is revealed.

PER-19A has been formally closed as Valid Full-Chain Qualification Evidence. A single case does not establish comprehensive platform qualification across indications, modalities, and design structures.

Where Does This Validation Model Sit Within the Current Industry Landscape?

In 2026, Onea conducted a Current-Industry Benchmark. Publicly available information shows that AI-assisted Study Design, Protocol generation, multi-model review, human oversight, and structured Protocol development already exist across the industry. These capabilities alone do not constitute Onea’s primary differentiation.

In the public materials reviewed, we did not identify the same complete combination of prospective evidence isolation, independent reasoning, design freeze, Sponsor-facing documents, chain-fidelity review, and delayed historical-comparator opening used in Onea’s end-to-end validation model.

Onea’s differentiation is not whether “AI can write a Protocol,” but how the Evidence-to-Protocol derivation chain is demonstrated to be trustworthy.

This comparison reflects publicly available information reviewed as of October 11, 2026. Non-public internal processes used by other organizations are outside the scope of this assessment.

Case 02 | Clinical Development Due Diligence

Platform Validation Extends Beyond Study Design and Protocol Development

Onea also tests whether AI-assisted analysis can identify critical issues, distinguish genuine development concerns from false positives, and produce traceable analytical outputs after human professional review within a Clinical Development DD workflow.

Using a blinded composite case, we evaluated whether AI-assisted Clinical Development DD could consistently identify known development issues while maintaining the boundary between development assessment and investment judgment. The test used a completed-case structure: AI-assisted analysis and human professional review were performed without access to the historical answer, followed by independent comparison with the historical result.

The objective was not to test whether AI could determine whether a company was “worth investing in.” It was to determine whether the workflow could identify material development issues, distinguish issues from stage-appropriate findings, connect them to relevant development milestones, and avoid crossing the boundary from clinical development assessment into investment recommendation.

This validation does not demonstrate that AI can make VC investment decisions. It tests whether AI-assisted analysis, combined with human professional judgment, can produce a traceable and development-relevant Clinical Development assessment without crossing investment-decision boundaries.

Validation scope: This case was an internally constructed composite scenario used to evaluate methodology and workflow. It does not represent an actual client engagement, investment outcome, or regulatory conclusion.

GxP-aligned validation principles do not imply formal GxP certification or qualification of the platform for all regulated uses.

Data Security & Confidentiality

More Than AI-Generated Protocols

General AI protocol generation starts with “How should this be written?” Onea starts with “What does the evidence actually support?”

Establish evidence-supported development decisions before moving into Study Design and document development.

Onea does not begin by asking AI to generate a complete-looking Protocol. Instead, it first controls and organizes the evidence, develops the Study Design based on what the evidence can support, keeps unsupported decisions OPEN, and only then proceeds to document development.

The Technology Platform Is Not a Standalone Product.
It Is the Foundation of How Onea Works.

The Onea AI Technology Platform is not offered as standalone software, a licensed product, or an automated decision-making tool. It is integrated into Onea’s clinical development advisory, study design, Pre-IND, IND, clinical writing, and due diligence workflows to expand analytical capacity, improve consistency, and support evidence-grounded human judgment.