AI & Clinical Documentation Policy

Effective Date: 28 July 2026

This Policy explains how Dealsparrow Limited, trading as NeuroClo (“NeuroClo”), provides AI-assisted features related to transcription, clinical documentation, and workflow assistance, and how responsibility is allocated between NeuroClo and Healthcare Providers (Customers).

It should be read with our Privacy Policy, Terms of Service, DPA, BAA, and Security Overview.

1. Scope

This Policy applies to optional or integrated features that may include:

• speech-to-text transcription (AWS-hosted, where enabled);
• clinical note assistance and draft documentation;
• document or encounter summarisation;
• search and retrieval assistance;
• workflow assistance; and
• other machine-assisted functionality using large language models or related AI services hosted by Amazon Web Services or other approved providers.

Certain features may not be available in all versions of the Services. Features involving audio, video, recording, transcription, or AI-assisted processing may be subject to additional terms, notices, or consent requirements.

2. Roles and responsibilities

NeuroClo

• Provides software tools that generate or suggest documentation-related output.
• Acts as a data processor / Business Associate (where a BAA is in effect) when processing customer/patient data for these features.
• Implements contractual, technical, and organisational safeguards with AI subprocessors.
• Does not practice medicine and does not make clinical decisions for patients.

Healthcare Provider (Customer)

• Remains the patient’s healthcare provider and controller (or equivalent) of clinical records.
• Decides whether to enable AI-assisted features for its organisation and users.
• Ensures lawful basis, notices, and consents (including for recording and transcription) under applicable law.
• Ensures clinicians review and approve AI-assisted content before clinical use.
• Maintains professional standards, documentation quality, and medical-record integrity.

Authorised Users (clinicians and staff)

• Use AI features only as trained and authorised by their organisation.
• Treat drafts, transcripts, and suggestions as unverified until reviewed.
• Correct errors before finalising notes or relying on summaries.

3. What AI output is — and is not

NeuroClo does not use personal data for solely automated decision-making that produces legal or similarly significant effects on individuals unless specifically disclosed and supported by an appropriate legal basis.

4. Data use and model training

• AI providers process data to deliver the contracted feature, under applicable laws, contracts, DPAs, BAAs, and customer instructions.
• Customer Personal Data and patient data are not used to train publicly available AI models unless expressly authorised by the Customer in writing.
• Whether a specific AWS or other provider feature retains data for service improvement under that provider’s HIPAA-eligible / contractual terms is addressed in product documentation and customer agreements; NeuroClo will not opt Customer into public-model training without express authorisation.
• Customers should review feature documentation before enabling transcription or generative documentation tools.

5. Recording, transcription, and telehealth

• Session recording, call recording, and transcription may be part of normal platform use where enabled.
• Healthcare Providers are responsible for informing patients and obtaining any required consents or authorisations under applicable law (including U.S., UK, and EU rules on recording and special-category health data).
• Where the platform supports consent or notice workflows, those mechanisms assist the provider; they do not transfer the provider’s legal responsibilities to NeuroClo.
• Transcripts and AI summaries may be stored separately from final clinician-authored notes, reflecting different lifecycles (raw/assisted artifacts versus approved clinical documentation).

6. Clinical documentation standards

Customers should ensure that organisational policies require, at minimum:

• human review of AI-assisted drafts before they become part of the official medical record;
• clear attribution or auditability of edits where the product supports it;
• correction of hallucinated, omitted, or misattributed content;
• no reliance on AI output for emergency triage or crisis decision-making; and
• training for Authorised Users on appropriate use and known limitations.

Emergencies: AI features must never be used as a substitute for emergency care. Seek local emergency services (911 / 999 / 112 as applicable).

7. Security and confidentiality

AI-assisted processing of PHI / health data is subject to the same security programme described in the Security Overview, including encryption, access control, audit logging, and vendor BAAs/DPAs for covered services. Subprocessors involved in AI features are listed on the Subprocessor List.

8. Patient portal and patient-facing content

Patients’ care relationship remains with their Healthcare Provider. Patient Portal Users receive access through their provider and may lose access if the provider suspends or terminates use of the Services. AI-assisted tools, if exposed in patient-facing contexts, remain subject to this Policy and do not create a clinician–patient relationship with NeuroClo.

9. Changes

NeuroClo may update this Policy as AI features evolve. Material changes may be notified via the website, email, in-product notice, or customer agreement channels.

10. Contact

• Email: hello@neuroclo.com
• Dealsparrow Limited, 395 Clapham Road, London, SW9 9BT, United Kingdom