Is your RIS-PACS AI-native, or just AI bolted on?

An AI-native RIS-PACS is one radiology platform where the RIS, PACS, reporting and AI share one architecture and one data layer. The AI shows up where the radiologist already works, not in a separate product joined by integrations. This page explains how that differs from a PACS with an AI plugin, the terms buyers come across, and what to check when a vendor uses the word.

The problem

• Software grew in layers. A RIS for registration and billing, a PACS for images, a reporting tool, sometimes Word. Then AI arrived as one more product, connected by an interface.
• Every join is a delay. A study waits between systems, a name gets retyped, or an AI result lands in a window nobody has open.
• Cloud alone doesn't fix it. Herman Oosterwijk notes in What's beyond PACS 3.0? that much cloud PACS is still the same software running in a virtual machine on someone else's server. "AI-native" is one answer: build the platform so intelligence is part of the workflow from the start.

AI-native vs bolt-on AI

	Bolt-on AI	AI-native
Where AI lives	A separate product talking to RIS and PACS	Inside the same platform as worklist, viewer and report
Logins	One per system	One
Patient data	Copied between systems, matched by ID	One record, one data layer
Adding an AI capability	Another integration project	A change in the same platform
Where the result appears	Often another window	In the viewer or report the radiologist already has open

Bolt-on AI has one real advantage: you can add it without replacing anything. Its cost is that every join is a place where a study waits, a name gets retyped or a result goes unseen.

Related terms

Unified RIS-PACS. RIS and PACS on one platform with one patient record and one login, from registration to signed report. A unified system is not necessarily AI-native; AI-native implies unified.

Agentic RIS-PACS. A RIS-PACS where software can carry out multi-step tasks on a user's request, such as finding a case or moving through the worklist, while the radiologist stays in control and signs every report.

Voice-controlled RIS-PACS. A system the radiologist can talk to: open the next case, navigate the worklist, dictate the report, hands-free.

Zero-footprint viewer. A DICOM viewer that runs in a web browser with nothing installed on the reading machine. It can render on the client or stream from a server.

What to look for

1. One path, one login. Ask to see a study go from registration and billing through the modality worklist, the viewer and the report to delivery, without retyping or a second login.
2. AI where the radiologist works. The AI result should appear in the worklist, viewer or report, not in another application.
3. The radiologist signs. AI-native is not autonomous reporting. Ask how drafts are reviewed and who signs.
4. Standards outward. DICOM for any modality, HL7/FHIR to the HIS, and full export if you leave.
5. Your own studies. Ask for the demo on a heavy CT from your own scanner, on your own connection.

How SOMATIQ handles it

Feature	What it means for you
Registration, scheduling, billing, acquisition, reporting and delivery on one platform	One patient record and one login from the front desk to the signed report
Our own zero-footprint viewer	3D/MPR, custom hanging protocols and mobile access, with nothing to install
Speech-to-text and LLM-assisted drafting	Reporting beside the images, with impression checks and inconsistency flags before sign-off
Voice-controlled worklist	Radiologists talk to the worklist and move through studies hands-free
HL7/FHIR integration	Connects to HIS, EMR and LIS

SOMATIQ is deployed across 130+ imaging centres in 3 countries.

Questions buyers ask

Is AI-native the same as AI reporting?

No. The radiologist reads every study and signs every report. AI removes work around the read.

Is a cloud PACS AI-native?

Not by itself. Cloud describes where software runs; AI-native describes how it is built.

Can an AI-native platform still use third-party AI?

It should, through standards such as DICOM and FHIR, so outside results reach the same worklist and viewer.

Do I have to replace everything at once?

That depends on your setup. Ask the vendor for a migration plan that keeps current studies and priors available from day one.

Further reading

• What "AI-native" RIS + PACS means
• RIS, PACS or RIS-PACS?
• What is a zero-footprint DICOM viewer?
• Talking to the worklist