Draft Capability & Proposal Narrative
BritiAI submission — NHS SBS Healthcare AI Solutions Framework (SBS10523)
This is drafting material — paragraphs to lift, adapt, and stitch into the Ariba form responses once the ITT structure is confirmed. Written to the win themes and to be tonally consistent across lots. Word counts indicative; tighten when ITT field limits are known.
0. Executive summary (single page — opens every lot response)
BritiAI is a UK-incorporated artificial intelligence company that exists to bring frontier AI into regulated environments responsibly. For this framework we have assembled a dedicated NHS delivery stack with two named partners: Votee AI, whose MAGIC LLM training platform and V-Note ambient capture tools are deployed across enterprise, government and financial services; and Beever AI, Votee’s peer-published research lab whose work on agentic systems and neural memory layers (ACL 2025; Vector Institute, Stanford, HKUST collaborations) gives every claim in this submission a citation rather than an assertion.
We have designed our offer around a simple principle: the NHS does not need another generic AI vendor — it needs an operating partner that treats clinical safety, data sovereignty and measurable outcomes as the starting line, not an afterthought. Every solution in this submission can be deployed inside a UK trust’s boundary, ships with a DCB0129-aligned Clinical Safety Case, and is anchored to a published 90-day KPI. Our advisory model is not a deck of slides — it is embedded operating capacity, including a Fractional Chief AI Officer service for trusts who need leadership before they need licenses.
We are bidding for Lots 5, 6 and 7 as primary, with Lots 3 and 4 retained as stretch decisions taken on 24 June. Lot 8 (Combined Solutions) is automatic for any supplier awarded a place on two or more lots, so requires no separate quality narrative; our integrated programme offer — Integrated Digital Front Door — becomes the call-off proposition once Lot 8 is conferred. We are deliberately not bidding for medical-device-class Lots 1 and 2 at this round; we will not overclaim.
1. About BritiAI (organisation section)
BritiAI was founded to translate horizontally-deployable AI capability into vertical, regulated, mission-critical use. Today we deliver seven productised AI capabilities — Social Listening, Copilot, Automated Speech Recognition, Image Detection, Document Processing, Report Generation & Business Intelligence, and a Data Processor — across finance, public sector and healthcare. We are UK-incorporated, UK-led, and operate with a fractional executive model that lets us scale delivery teams up and down with demand without compromising senior accountability.
For this framework we are the prime contractor. We hold the customer relationship, the contracting line, the clinical safety responsibility, and the day-to-day delivery accountability. Our partners are named, contracted, and bound to us under a binding Teaming Agreement that flows down all NHS data protection obligations.
2. About our partners
Votee AI is the technology partner. Votee operates a 95%-agentic LLM training pipeline (MAGIC) that allows enterprise clients to train models on proprietary data without data egress. Votee’s V-Note product — meeting transcription with on-premises deployment — is the basis of our ambient clinical documentation offer. Votee’s Chief Scientist is the former Dean of the Deloitte AI Institute, and the company holds the Baidu Innovation Award 2026 and was selected for the HKMA FSS 3.1 pilot programme.
Beever AI is the research partner. Beever is Votee’s research laboratory, with active collaborations with HKUST, Vector Institute, Stanford, the University of Waterloo and HKU. Beever’s published work — including TheoremExplainAgent and HKCanto-Eval (both ACL 2025) and the open-source Beever Atlas neural memory layer — gives this submission a rare property in the AI vendor market: every model behaviour we claim is benchmarked, and every benchmark is published.
3. Our delivery model (cross-lot)
We deliver against a four-stage model that NHS trust SROs recognise from any major change programme:
- Diagnose (Weeks 1–2). DTAC-aligned readiness assessment, data flow mapping, use-case prioritisation against trust strategic objectives.
- Design (Weeks 3–4). Solution architecture, Clinical Safety Case kick-off (DCB0129), DPIA, success metrics agreed in writing with the SRO.
- Deploy (Weeks 5–10). Iterative, sprint-based deployment with a named on-site delivery lead, weekly clinical-engagement sessions, and twice-weekly executive read-outs.
- Demonstrate (Weeks 11–12). Published KPI report against the agreed metric, transition to BAU, and an optional embedded operating partner under our Lot 7 advisory.
LOT 6 — Operational Efficiency (PRIMARY BID — quality-lead)
Award criteria: Quality 90% / Social Value 10%. No cost weighting. This is a pure-quality race — every paragraph below is written to differentiate, not to undercut.
6.1 Why we win this lot
The single most-cited frustration in NHS clinical workforce surveys is administrative load. Yet the dominant ambient-documentation vendors today route audio through US cloud infrastructure, struggle with DSPT-compliance edge cases, and price per-clinician at levels that exceed many trusts’ digital budgets entirely.
We offer a different model. Our Scribe-On-Site solution runs ambient capture and summarisation on the trust’s own infrastructure (Votee V-Note on-prem), wraps it with the BritiAI Copilot for note structuring and coding suggestions, and is priced at a level designed to be affordable for routine deployment across an entire directorate, not a handful of pilot clinicians.
6.2 Named solutions under Lot 6
(i) Scribe-On-Site — Ambient Clinical Documentation On-prem ambient capture + summarisation + structured note output, fully aligned to DCB0129 and DTAC v2. Target KPI: ≥60 minutes per clinician per day of documentation time reclaimed. Deployment: 8 weeks from call-off.
(ii) WardFlow Copilot — Agentic Discharge Orchestration Agentic workflow drafts discharge summaries from the EHR, books TTOs, flags blockers to the discharge co-ordinator, and routes to the responsible consultant for sign-off. Target KPI: 25% reduction in time-to-discharge-letter; 10% improvement in TTOC compliance.
(iii) Atlas for Trusts — Neural Memory Layer Beever Atlas deployed as a trust-wide retrieval and reasoning layer over policies, SOPs, clinical guidelines, M&M notes and learning logs. Target KPI: ≥40% reduction in time-to-find for clinical and operational queries; full audit trail on every answer.
(iv) FOI & Complaints Auto-Triage High-volume, low-clinical-risk entry point. Documents are classified, summarised, routed and partially drafted, with human-in-the-loop sign-off. Target KPI: 50% reduction in median triage time; zero increase in upheld complaints.
6.3 Evidence
- Votee V-Note: production deployments at [named enterprise references — to populate from Votee]
- Beever Atlas: open-source release with adopter community; benchmarked retrieval performance against [comparator] published in [reference]
- BritiAI Document Processing & Copilot: deployed in insurance claims and legal document review (case studies attached)
- Clinical Safety: DCB0129-aligned Clinical Safety Case template, hazard log, and named CSO appointed for the framework
6.4 KPIs we commit to in writing
| Solution | KPI | Measurement window |
|---|---|---|
| Scribe-On-Site | ≥60 min/clinician/day reclaimed | 90 days from go-live |
| WardFlow Copilot | ≥25% reduction in discharge letter cycle time | 90 days |
| Atlas for Trusts | ≥40% reduction in median time-to-find | 90 days |
| FOI Triage | ≥50% reduction in median triage time | 60 days |
6.5 Pricing model
Per-user/month subscription for Scribe-On-Site and Copilot tools; flat trust-wide licence + implementation fee for Atlas and FOI Triage. Volume discounts at 5/10/25 trust thresholds. Outcome-linked pricing variant available where the trust agrees a baseline measurement.
LOT 7 — Advisory & Specialised Support (PRIMARY BID — pricing matters)
Award criteria: Quality 70% / Commercial 20% / Social Value 10%. The 20% commercial weighting means our rate card must defend itself against Big-4 (more expensive) and boutique specialists (often cheaper). Mid-market premium with transparent value framing is our position.
7.1 Why we win this lot
Most NHS AI advisory engagements deliver a strategy deck and a stage-gate plan, then leave. Trusts then discover the gap between strategy and execution is exactly where the work — and the risk — lives. We close that gap by offering embedded operating partnership, not consultancy days.
7.2 Service catalogue
- AI Readiness Programme — 12-week embedded engagement: DTAC v2 gap analysis, use-case prioritisation, data architecture review, Clinical Safety governance set-up, executive coaching, board reporting. Fixed price; outputs include a board-approved AI strategy, a 12-month deployment roadmap, and a named operating partner relationship.
- Fractional Chief AI Officer — Senior leader, embedded 1–3 days/week, accountable for the trust’s AI portfolio. Typically engaged for 6–18 months. Designed for trusts that need leadership before they need licenses.
- Procurement & Vendor Assurance — Independent technical and clinical-safety assurance for trusts evaluating third-party AI vendors under this framework or others.
- Workforce Enablement — Role-based AI literacy training (Board, executive, clinical leadership, frontline), aligned to NHS England’s AI knowledge repository.
- Implementation Pods — Mixed BritiAI + trust staff delivery teams for specific AI deployments, working sprint-based with a named on-site lead.
7.3 Day rate card
| Grade | Day rate (£) |
|---|---|
| Analyst | 550 |
| Consultant | 750 |
| Senior Consultant | 950 |
| Principal | 1,450 |
| Fractional Executive (CAIO/CTO/COO) | 2,200 |
All rates subject to standard framework discounting on volume. Rates fixed for 24 months from call-off.
7.4 Evidence
Fractional executive model is BritiAI’s operating norm. Provide CVs of the named delivery cadre; cite Beever’s research credentials as the technical backstop. Reference letters from existing BritiAI clients (non-NHS) demonstrating embedded delivery model effectiveness.
LOT 5 — Research, Innovation and Development (PRIMARY BID)
Award criteria: Full weighting to be confirmed from procurement documents.
5.1 Why we win this lot
This lot rewards what Beever already does. We bring an active research lab with peer-reviewed work at ACL 2025 and collaborations with Stanford, the Vector Institute and HKUST. We are not learning research process for the first time.
5.2 Named solutions
(i) MAGIC-NHS — Sovereign LLM Training Pipeline A managed environment letting NHS organisations fine-tune large language models on their own clinical, operational and research data without that data leaving the trust boundary. Pipeline includes data preparation, training, evaluation against published benchmarks, governance documentation, and a model card.
(ii) Trial Concierge — Agentic Clinical Trial Recruitment & Support Agentic system that supports trial teams with patient eligibility screening (from EHR signals, with clinical sign-off), protocol Q&A for clinicians, and recruitment letter drafting. Designed to plug into existing CRF/EDC systems.
(iii) Cross-Trust Federated Research Substrate Beever Atlas deployed federated across multiple trusts to enable collaborative research workspaces while keeping each trust’s underlying data in place. Built for NHS Research Ethics-compatible governance.
5.3 Evidence
- ACL 2025 publications (TheoremExplainAgent; HKCanto-Eval / CoNLL)
- Open-source release of Beever Atlas (with adopter metrics)
- MAGIC training pipeline production deployments at [named enterprise / public sector references]
- Active university partnerships listed above
LOT 8 — Combined Solutions (AUTOMATIC — no separate bid)
Per the published notice: “Suppliers awarded a place on a minimum of two separate Lots will automatically be eligible to offer a Combined Solution to Approved Organisations via Lot 8, ideally to provide an ‘end to end’ process.”
No standalone Lot 8 quality response is required at submission. Enrolment follows from winning Lots 5, 6 and/or 7.
Our Lot 8 call-off proposition — Integrated Digital Front Door
Once enrolled in Lot 8, we will offer Approved Organisations a multi-year programme combining Scribe-On-Site, WardFlow Copilot, Atlas for Trusts and the Fractional CAIO advisory wrap. Single trust SRO, single delivery lead from BritiAI, single quarterly executive review. Designed for trusts that are committing to AI as an operating model, not a pilot. Optional outcome-share pricing: a portion of fees is held back and released against documented achievement of pre-agreed KPIs.
This narrative is preserved here for use in post-award trust call-off pitches, not the framework submission itself.
CROSS-CUTTING SECTIONS
Clinical Safety (DCB0129)
We operate under a manufacturer Clinical Safety Management System aligned to DCB0129:2018. A named Clinical Safety Officer is appointed for this framework. Every solution ships with a Clinical Safety Case, a Hazard Log, and a Clinical Risk Management Plan. We engage with each deploying trust’s DCB0160 process and provide the documentation the trust’s CSO needs to discharge their assurance role.
Information Governance & Cyber
- ISO 27001:2022 [status to confirm — Stage 1 in flight; roadmap to certification by Q4 2026]
- Cyber Essentials Plus [scheduled for assessment by 18 June 2026]
- NHS Data Security and Protection Toolkit: registered, working toward Standards Met
- UK GDPR / DPA 2018 compliant; ICO registered
- Data residency: UK-only inference and storage; on-prem option available for every productised solution; named UK sub-processors only
- DPA flow-down to Votee and Beever as named sub-processors under binding Teaming Agreement
Sustainability & Social Value
- Carbon-aware inference scheduling; UK-grid-aware deployment topology
- Apprenticeships and placements through BritiAI’s UK delivery cadre
- Open-source contributions (Beever Atlas) as social value commitment
- Commitment to publish a Social Value Plan annually under each call-off
Equality, Diversity & Inclusion
- EDI policy attached; supplier diversity track record demonstrable through partner selection
- Algorithmic bias testing built into every model release cycle; bias evaluation reports shared with each deploying trust
APPENDIX A — Frequently Asked Buyer Questions (pre-empted)
Q. Can you scale beyond one trust? A. Yes. Our productised solutions are designed for multi-tenant deployment with single-tenant data boundaries. Our advisory model scales by adding named principals from our delivery cadre; we have a recruitment pipeline ready to activate on framework award.
Q. What happens if Votee or Beever changes ownership or fails? A. Our Teaming Agreement contains step-in rights and source-code escrow. In the event of partner failure, BritiAI retains the right to continue delivery under the framework using the licensed IP, with a transition window agreed with the buying authority.
Q. Where is the data? A. Inside the UK, and for our productised solutions, inside the trust where you want it. On-prem deployment is supported for every Lot 6 solution.
Q. Are you an AI medical device? A. Our current Lot 5, 6, 7 and 8 solutions are deliberately scoped outside Class IIa medical device territory. We do not make diagnostic or treatment decisions; we support clinicians and operational teams. Should a future call-off require a medical-device claim, we will follow the MHRA software-as-a-medical-device route in full before commencing that work.