Everything you need to know

Care Continuo AI reads incoming hospital letters in your SystmOne Documents Inbox, automatically extracts every follow-up action they contain including blood tests, medication changes and referrals. It then presents them as a structured list for your review in the Scan Action screen. You approve or reject each suggestion; nothing is actioned without your explicit decision. It replaces the manual task of reading each letter and mentally constructing a to-do list.

Currently, Care Continuo AI is focused on hospital correspondence. This includes discharge summaries, outpatient letters and post-operative letters. It extracts four action types: blood test requests, medication changes, referrals (with AI-drafted referral letters), and recalls. We are deliberately narrow at this stage to ensure excellent performance in this specific context before expanding further.

No, Care Continuo AI is an administrative AI tool. It extracts stated follow-up instructions from letters. It does not interpret clinical results, generate diagnoses or make clinical recommendations. The decision about whether each suggested action is appropriate for a specific patient always rests with the reviewing clinician. This positioning is deliberate and is central to our clinical safety framework and regulatory approach.

Each extracted task is shown with a confidence percentage. Tasks with ≥90% confidence are shown in green; lower-confidence tasks are flagged in amber, prompting the reviewer to apply extra scrutiny. The confidence score reflects the model's certainty that the identified action matches what the letter is requesting. It is always visible before you make an approval decision, so you can weight it accordingly. We will publish accuracy and calibration data from our validation studies before any clinical deployment.

The tool adds a new "AI Suggested Actions" section to the existing SystmOne Scan Action screen (the screen GPs already use to file every incoming letter). The goal is that the tool feels like a natural SystmOne feature, not an external application, using the same chrome, fonts, and visual language as SystmOne itself.

EMIS integration is on our roadmap. We are focusing on SystmOne first given that it is the most widely used clinical system in our initial target market, but we are actively investigating the EMIS developer programme in parallel. If you're an EMIS practice interested in early access, please get in touch - your input will directly shape our integration priorities.

The workflow changes only at the Scan Action screen i.e. the final step before a letter is filed. Everything up to that point (letter arriving in the inbox, opening it, reviewing the document details etc) is identical to what you do today. The AI panel appears in the existing left-hand sidebar of the Scan Action screen, alongside the existing Read Codes section, so it requires no new habit or additional navigation.

The governance gate is a design decision that disables the "Complete" option in the Scan Action screen until every AI suggestion has been explicitly approved or rejected. This means it is physically impossible to file a letter with unreviewed AI suggestions still pending. This operationalises human-in-the-loop oversight at exactly the right moment - before a letter is filed to the patient record. We believe this is a meaningful differentiator for NHS clinical safety reviewers evaluating tools under DCB0129 and DCB0160.

Care Continuo AI is being designed with DCB0129 (Clinical Risk Management: its Application in the Manufacture of Health IT Systems) and DCB0160 (Clinical Risk Management: its Application in the Deployment and Use of Health IT Systems) compliance in mind from the outset. The audit trail, governance gate, and human-in-the-loop architecture are all direct responses to the requirements these standards place on AI-assisted clinical tools. We will publish our Clinical Safety Case as part of our validation programme.

Patient data security is a non-negotiable requirement. All data processing will occur within UK data centres compliant with NHS DSPT (Data Security and Protection Toolkit) requirements. We are committed to NHS-compliant data handling, including data minimisation, appropriate retention policies, and full transparency with practices about what data is processed and how. Our data processing agreement will be available before any clinical deployment. We will never use patient data to train AI models without explicit consent and ethical approval.

We are currently in a pre-pilot phase, working with a small number of forward-thinking practices to validate the product and refine the integration. If you'd like to be considered for early access, please get in touch via our Contact page. We're particularly interested in hearing from practices with high correspondence volumes or a strong interest in clinical AI governance.

Pricing is not yet finalised and will be shaped by what we learn during the pilot phase. Our expectation is a per-practice monthly subscription model, with pricing designed to be achievable for individual practices as well as attractive at PCN-level group purchasing. Early access participants will receive preferential pricing. We will not charge for the initial pilot period.