LegalDocs-Assist is a proprietary SaaS platform purpose-built for UK legal practitioners working in RTA credit hire, housing disrepair, and clinical negligence. It combines enterprise-grade AI with a curated legal knowledge base to help teams analyse correspondence, draft responses, generate witness statements, and manage cases — all with industry-leading safeguards against AI errors.
Three-page wizard that drafts a CPR Part 32-compliant witness statement from structured inputs, ready for review, statement of truth and signing.
Paste or upload insurer or defendant correspondence. The AI identifies cited case law, validates it against our verified database, and highlights unsupported arguments.
Draft professional, citation-backed responses grounded in verified precedent. Customisable tone and strategy — firm, conciliatory, or without-prejudice.
Evaluate an accident statement against authorities and receive an indicative liability apportionment, with reasoning grounded in the curated case law set.
210 UK authorities, searchable and categorised. Each entry includes binding/persuasive status, key principles, and common insurer/defendant arguments addressed.
Curated hub of GTA protocols and rate tables, ABI guidance, Law Society and MoJ procedural rules, Road Traffic Act statutory references, and major hire-provider T&Cs.
Combines, deduplicates and paginates multi-file medical record sets into a single court-ready bundle, with an automated dedupe report.
Builds a date-ordered chronology of disrepair events, outstanding defects and severity flags from tenant correspondence and inspection records, exportable to Word or CSV.
Our AI layer is powered by Google Gemini 3 — one of the most capable large language models available — but wrapped in our own proprietary safeguards:
Additional safeguards include a postcode fabrication scrubber (prevents the AI introducing postcodes that did not appear in the original input — postcodes are deliberately preserved through the masking gateway as material to legal locality arguments) and BHR vehicle challenge verification against known provider databases.
🔒 Source documents stay on the user's device — text‑based PDFs and Word files are processed in the browser; only the extracted text is sent to our server. Scanned/image‑only PDFs are briefly uploaded for OCR and deleted within the hour.
🛡️ PII masking gateway — a mandatory server‑side gateway detects and replaces personal identifiers (titled names, emails, UK phone numbers, NI and NHS numbers, driving licences, VRMs, bank details, DOBs, references and street addresses) before any text reaches the AI. It is a strong first layer but pattern‑based, not a perfect filter — send only what the task needs. UK postcodes are deliberately preserved (material to BHR locality and venue analysis) and protected by an output‑side anti‑fabrication scrubber.
🤖 AI runs on our server, under Zero Data Retention — drafting and generation happen server‑side; the provider does not retain or train on the data; the generated output is returned to the user and then deleted, not retained.
♻️ Designed not to retain case data — uploaded source documents and generated outputs are deleted shortly after they are produced or on session close, with a continuous backstop purge so nothing is held beyond 24 hours. Firms download or save their own copy and re‑run if needed. Uploaded files are never included in any backup.
🩺 Special‑category (health) data — the Service routinely processes medical records for clinical‑negligence and personal‑injury work; the safeguards above (browser‑side processing, server‑side masking, encryption, immediate deletion) apply. As controller, the firm is responsible for the Article 9 lawful basis.
🏢 Row‑level security — every company's data is isolated at the database level. No cross‑client access is possible.
🚫 No training on your data — your correspondence is never used to train AI models.
🔐 Encrypted at rest & in transit — all data protected with industry‑standard encryption.
Enterprise cloud infrastructure with automatic scaling, built-in redundancy, and high availability. No on-premise installation required.
Fully managed, encrypted relational database with row-level security policies enforcing strict data isolation between clients.
Backend logic runs on serverless edge functions — automatically scaling with demand. No cold starts, no capacity planning.
Responsive React web application accessible from any device. No software installation — works in any modern browser.
✅ GDPR compliant — data minimisation, privacy by design, and full transparency.
✅ ICO registered — Registration number ZC013423.
✅ Subject access rights — users can request access to, correction of, or deletion of their data at any time.
✅ AI provider terms — Google's enterprise terms prohibit use of customer data for model training.