Faster certification does not, by itself, establish either weaker scrutiny or better assessment. The relevant question is what became faster: waiting for a reviewer, organising evidence, resolving questions or generating the evidence itself. Without comparable starting points and scopes, a headline timeline cannot tell a clinical buyer how thoroughly a device was assessed.
In its 12 October 2026 funding announcement, Scarlet says customers have cut a year or more from new-device certification timelines and reduced update timelines from months to weeks. These are company claims, not independent comparative findings. They provide a reason to examine the process, not a basis for declaring every faster assessment either safer or more permissive.
There is more than one clock
Imagine two fictional manufacturers reporting the time to certification. One starts its clock when it first contacts an assessor, before completing its clinical evaluation. The other starts when a submission has passed an initial completeness review. Their reported durations describe different journeys.
A third may exclude periods when the manufacturer is responding to questions. That measure can be useful for assessing reviewer responsiveness, but it is not the same as the total time a founder must budget before a decision.
A proposed timeline record should therefore separate initial enquiry, eligibility review, application acceptance, active assessment, manufacturer responses and final decision. The aim is not to select the shortest number. It is to understand which delays are avoidable and which reflect necessary work.
Faster because of what?
The following framework is an editorial tool for investigating speed claims. It does not describe measured Scarlet performance or imply that any particular practice occurs there.
| Possible explanation | Why it could be legitimate | What to check |
|---|---|---|
| Better document organisation | Reviewers spend less time finding the relevant evidence | Whether traceability and substantive review remain intact |
| Earlier scope clarification | Fewer avoidable misunderstandings reach formal assessment | Whether the same intended purpose is being compared |
| More relevant reviewer capacity | Suitable applications wait less for allocation | Whether expertise matches the specific device |
| More complete submissions | Fewer requests are needed for missing material | Whether manufacturer preparation is included in the timeline |
| A narrower product claim | Less evidence may be relevant to the bounded use | Whether the resulting scope is narrower too |
| Reduced substantive examination | Time falls for the wrong reason | Whether requirements, findings and decision rationale are adequately addressed |
The final possibility deserves scrutiny, but should not be inferred merely because a process is efficient. Equally, administrative sophistication should not be accepted as proof that clinical evidence has received adequate examination.
Evidence generation is not an administrative delay
A fictional start-up wants to expand software from a specialist population to general practice. Its existing study may not answer questions about the new patients, users or workflow. Compressing the document-review stage does not generate that missing evidence.
The MHRA's intended-purpose guidance, checked on 13 October 2026, connects the claimed use to the supporting evidence. A shorter assessment of a restricted indication cannot be presented as equivalent to a broader assessment unless the scope genuinely matches.
This is where a founder's commercial timetable can become misleading. A promised launch date may assume that no important gap will be found. A responsible assessment process must still be able to identify a gap that changes the timetable or prevents the proposed claim.
Compare equivalent engagements rather than brand averages
A useful comparison would control for device type, risk, clinical novelty, dependence on external models, submission readiness and the assessment route. It would also account for applications that were withdrawn, refused or required substantial additional work.
Our proposed evaluation would report the distribution of durations rather than only a best-case example. It would explain what proportion of delay lay with the assessor, the manufacturer and external evidence generation. These are recommendations for transparency, not published results from a head-to-head certification study.
A buyer should also ask what happens after the first decision. A fast initial engagement with unclear change review, limited communication or poor continuity may not be the most useful arrangement for a product that changes regularly. Conversely, a longer initial assessment may establish a clearer foundation for later bounded updates.
Automation can assist review without replacing judgement
In its 12 October 2026 investment account, Headline describes a thesis in which technology supports evidence organisation while qualified people make regulatory judgements. That is the investor's account of the proposition, not independent proof of review quality.
The important assessment question is how an organisation verifies the work performed by its own tools. If software identifies apparently missing evidence, can the reviewer inspect the basis? If it summarises a study, are limitations preserved? If it suggests that a requirement is addressed, does someone establish whether the evidence actually supports the claim?
These are sector-wide governance questions. They apply whether the organisation is venture-backed, established, highly automated or largely manual. An assessor's own technology should not become an invisible source of unsupported certainty.
Appropriate scrutiny includes the ability to say no
The MHRA's description of approved-body oversight emphasises competence, impartiality and objective assessment, together with continuing monitoring after designation. Those characteristics matter more than a promise that every customer will reach a favourable decision quickly.
A founder can reasonably expect clear communication, predictable processes and an explanation of findings. They should not expect certainty of certification before the evidence has been assessed. A contractual or marketing proposition that confuses responsiveness with guaranteed acceptance deserves careful examination.
The same applies to updates. The joint change-control principles concern specified modifications and their supporting controls. A faster route for a bounded, adequately assessed change is not a general permission to alter clinical purpose or behaviour without further consideration.
A practical way to prepare without lowering the standard
For the fictional manufacturer, the most productive preparation would be to settle the intended purpose, map each claim to evidence, identify unresolved gaps and organise the materials so a reviewer can follow the argument. Contradictions should be exposed before submission rather than hidden in separate documents.
As described on 13 October 2026, Insights by iatroX offers clinical safety and readiness support. The relevant advisory contribution is clearer evidence and ownership, not a promise to influence an independent assessor's conclusion.
For a submission-ready team facing allocation delays, additional capacity may be valuable. For a team whose evidence is incomplete, better preparation matters more than a shorter advertised clock. For an NHS buyer, the resulting scope, evidence and deployment assumptions remain more important than how many weeks the manufacturer says the process took.
Frequently asked questions
Does fast certification mean an AI device received less scrutiny?
Not necessarily. Faster organisation, better preparation or greater reviewer capacity can reduce delays without changing the substantive requirements.
How should founders compare certification timelines?
Use equivalent assessment scopes and clear starting and stopping points. Include manufacturer response time, evidence gaps and the treatment of unsuccessful or withdrawn applications.
Can a faster assessment compensate for missing clinical evidence?
No. A more efficient process can identify and assess evidence, but cannot establish results that have not been generated for the relevant claim.