Clinical Audit · Pilot

On-premises AI clinical audit.

Audits that used to take an afternoon of two systems side by side and a spreadsheet, done end to end on the hospital's own servers. Every finding carries its evidence, every decision carries a name, and nothing is changed in a clinical system on Oriel's own.

Book a demo Watch the 4-minute demo

Audit leads, department managers, clinical governance and quality teams, and the IT and information-governance staff who deploy for them, starting with the cath lab and imaging and extending to pathology, pharmacy, research and any department that audits against a standard.

The problem

Every department runs audits: device coding, radiation dose, missing images, turnaround times, controlled drugs, consent. Most are still done by hand by reading one system against another, so they are slow, sampled rather than complete, and hard to repeat the same way next month.

How it works

1
Ask
Describe the audit in plain words, or pick a pack from the library.
2
Approve the plan
See what Oriel will read and which standard applies, then approve it, edit it or cancel.
3
Decide the findings
Review each case with its evidence and agree or change it, with a note.
4
Sign off and repeat
Confirm the drafted report, then save the audit as a schedule or workflow for next month.

Ask in plain words what you want to check, for example September's pacemaker exams in the RIS against the implant images in PACS. Oriel writes a plan saying what it will read, where, and which part of the local procedures applies. Nothing runs until a person approves it.
Once approved, it works through every case using standards-based connectors (FHIR, DICOMweb, HL7 v2) or, where a system has no interface, on screen through a workstation runner that masks patient names and stops the moment staff touch the mouse. The canvas fills with the evidence as it goes.
Each finding arrives with its evidence and a reason. A person agrees or changes it with a note. Oriel then drafts the audit report in the department's own template, cited to the guidance and the cases, for sign-off. The audit can be saved as a versioned pack, a schedule or a workflow and run again.

See it working

A pacemaker coding audit from start to finish: ask in plain words, approve the plan, review every case with its evidence, then guidance, computer use, the report, sign-off, workflows and support. Fictional demo hospital, synthetic data.

Inside Oriel

Oriel thread for a pacemaker coding audit showing a five-step plan waiting for approval, with the employer's procedures open in the canvas — The plan — asked to check September's pacemaker exams against the implant images, Oriel lists what it will read and where, and cites the part of the local procedures that applies. Nothing runs until a person approves it. Fictional demo hospital, synthetic data.

Oriel reviewing a case: a fluoroscopy frame of an implant, the recorded code, Oriel's suggestion, and Agree or Change buttons — Review — each case shows the images, what was recorded and what Oriel thinks the device is, with its reason. A person agrees or changes it; nothing is written back to the radiology system.

Oriel audit dashboard: 64 cases checked, 8 wrong codes, recorded versus observed device types, and the activity log — Results — every exam checked rather than a sample, with recorded against observed device types, the cases that differ, and who decided each one.

Oriel drafting an audit report with background, standards and methodology sections, each citing guidance and cases — Report — drafted in the department's own template, with citations to the guidance and to every case, ready to review section by section, export and sign off.

Oriel drafting a new audit pack for controlled drug witnessing in pharmacy from a one-line description — Any department — a pharmacy lead describes a controlled drug witnessing check in one line, and Oriel drafts it as an audit pack for governance to approve. Packs are data, not code.

Oriel workflow editor with a start trigger, a system check step and a notify step — Workflows — chain checks, audits, approvals and notifications, and start them by hand, on a schedule, from a webhook or when an audit finishes.

What it does

Plain-language audits with a plan to approve before anything runs, and pause, stop or fork at any step
Evidence on every finding, decided by a named person; Oriel never corrects a record on its own
Audit packs as versioned data, drafted from a description and approved by governance, plus a standards library
Guidance answers cited to the section and page of the Trust's own documents
Connectors for FHIR R4, DICOMweb PACS and HL7 v2 over MLLP with TLS
Computer use on workstations for systems without an interface, with masked screenshots and a Stop banner
Report drafting in the department's own DOCX or PDF template, with export and sign-off
Schedules and workflows triggered by hand, by a webhook, on a timetable or when an audit finishes, with email and in-app notifications
Models run on site and every step has a non-AI fallback; role-based access and a full action log
Desktop app for one PC, or a server install with encrypted backups, readiness checks and a scrubbed support report

Questions

Does Oriel make the audit decisions?

No. It proposes a plan for a person to approve, gathers the evidence, and suggests a finding for each case. A named person agrees or changes every finding and signs off the report. Anything that would write to a clinical system needs explicit approval from someone allowed to make that change.

Where does the data go?

Nowhere outside the hospital by default. Oriel runs on the hospital's own servers or a desktop in the department, and its models run on site. A hosted model can be added only if the site chooses to, and patient identifiers are pseudonymised before any outside call.

Which departments is it for?

Any department that audits against a standard. The demo hospital includes a cardiac catheter lab (pacemaker device coding, radiation dose), imaging (exams without archived images), pathology (turnaround breaches), pharmacy (controlled drug witnessing) and research. New audits are written as packs, not code.

What if a system has no interface?

Oriel can work on screen the way staff do. A small runner on the workstation lets it read the department's own program; names are blacked out before any screenshot is kept, a banner shows what it is doing, and pressing Stop or moving the mouse pauses it at once.

Can I see it?

Yes. Watch the demo on this page, or register for the pilot and we will run it with you on synthetic data and talk through what a departmental pilot would need.

Try Oriel in your department

Oriel is in pilot. Book a demo and we'll show it on synthetic data, then talk about running it on your own systems, under your governance.

Book a demo

Or email hi@zapxlabs.com.

Oriel is a working prototype shown on a fictional demo hospital with synthetic data only. It is not a medical device and is not for clinical use. It drafts and flags for people to decide; it does not diagnose or change clinical records. A clinical safety case and information-governance approval come before any use with real patient data. Oriel is a working name.