Built, working, demonstrable

The Medical Oversight Workbench

It sits beside your stack — not your EDC, not your safety database, not a rip-and-replace. It ingests what your systems already produce and adds the one thing they cannot: documented physician judgment, bound to every decision.

What it does

Physician judgment, at signal speed

Narratives at signal speed

SAE narratives drafted in seconds — reviewed, edited, and signed by a physician. Never auto-filed.

Grading & triage support

AE grading assistance and signal triage across your program — the physician always in the loop.

Sign-off that binds

Part 11-aligned e-signatures — identity, timestamp, and meaning bound to every decision.

🛡

An audit trail by default

Tamper-evident record of every AI interaction and clinical call — inspection-ready as work happens.

It sits beside your stack

Not your EDC. Not your safety database. Not a rip-and-replace.

The Workbench ingests what your systems already produce and adds the one thing they cannot: documented physician judgment.

Live demo, now: eight minutes — a case like yours, end to end.

Book the demo →
The architectural invariant

We built the bypass out

No AI output becomes a record of decision without a physician's review and signature. The regulatory obligation is human — so the architecture makes the human unavoidable.

1
Signal arrives

AI-flagged event lands in the review queue with full source data.

2
AI assembles

Draft narrative and suggested grade in seconds — labeled suggested.

3
Physician decides

Named reviewer adjudicates — can override, documents rationale.

4
Sign-off binds

Part 11-style e-signature: identity, time, and meaning locked in.

5
Trail exists

Tamper-evident audit record, generated as the work happens.

The invariant, on screen

A flag is a suggestion. The signature is the decision.

This is a real case in the Workbench today. The AI assembles the analysis, causality, and grade — every line labelled suggested. Nothing becomes a record until a named physician adjudicates and signs.

  • AI analysis with explicit confidence — never auto-filed
  • Causality and expectedness surfaced for the reviewer to confirm or override
  • Confirm & Sign — the Part 11-aligned moment the decision binds
Workbench case view: AI analysis with causality and expectedness, awaiting physician Confirm & Sign
The audit trail exists by default — not by request

Every decision is the record.

Every physician action — confirmation, override, or rejection — is captured as it happens. No separate logging step. No reconstruction from memory. The trail is the byproduct of doing the work.

Physician override modal: causality, seriousness, and expectedness with a documented rationale field — the reason becomes part of the immutable record
Override modal — the physician documents the rationale for every change. The reason is the record.
Audit Trail Report: CFR 21 Part 11-aligned log with timestamps, reviewer identity, changed fields, and documented rationale — exportable for inspection
Audit Trail Report — who, when, what changed, and why. CFR 21 Part 11-aligned, exportable for inspection.
Where it's going

A compliant, AI-driven unified platform

The Workbench is the first module of a unified, GxP-validated environment that will bring signal detection, medical oversight, and the documented decision record into one compliant system of record — built on the same architectural invariant: the physician decides, and the trail proves it. It is the natural extension of the industry's first unified eClinical platform, now purpose-built for the RTCT era.

Today

The Workbench — physician oversight beside your existing stack.

Next

Native signal detection and case intake under the same compliant decision layer.

The platform

One unified, validated system of record for AI-assisted, real-time clinical safety.

Where it's going — already taking shape

Signal detection, under the same decision layer

The native detection and case-intake modules on the roadmap are further along than "next" suggests. Here they are in progress — disproportionality and trend analytics that feed the same physician-owned, audit-ready sign-off.

Signal summary report: portfolio escalations, drug-event pairs, and disproportionality statistics
Signal summary — portfolio escalations and disproportionality (PRR/ROR) across the program.
Signal detail view: cumulative incidence trend and AI signal assessment with statistical summary
Signal detail — cumulative-incidence trend and AI assessment, with the statistical summary behind it.

Same invariant, one module earlier: detection assists, the physician still decides, and the trail still proves it.

See it on a case like yours

Eight minutes, end to end.

The fastest way to understand the Workbench is to watch a signal travel from AI flag to signed, audit-ready decision — on a case that looks like your program.

Book a live demo