BAiSICS
Enterprise AI platform that transforms legal document review through advanced machine learning and natural language processing.
Four hours of manual review per lease, with no AI tool firms could commit to.
Commercial property lease analysis is high-stakes, manual work. Documents are non-standardised, often skewed and poorly scanned, and amendments are frequently handwritten. Skilled professionals were spending whole afternoons on a single lease, and a misread clause carried material risk.
Ctrl+F could not handle the scan quality. Generic chat models could not be cited against the source. GPT-4 and the leading legal-AI tools on the market trimmed the work but failed the accuracy bar firms needed to commit to a position. None were production-viable.
- Up to four hours of manual review per lease.
- Source PDFs are skewed, low-quality scans with handwritten amendments.
- Standard OCR and generic LLMs cannot extract nuanced legal data accurately.
- Every output has to be verifiable against the original PDF, with no hallucinated clauses.
Inside the BAiSICS workspace.
Every analysed document, traceable by tag and client
Intelligent tagging classifies leases, titles, and deeds of variation automatically. Search across hundreds of documents in seconds, then export, edit, or hand off in one click.
Every answer linked to a paragraph in the source PDF
The original lease on the left with the relevant clauses highlighted. The extracted field on the right with its citation and confidence. Low-confidence answers are flagged for human review, never silently committed.
Firm-owned templates next to a curated standard set
Each firm keeps its own templates next to a standard library for common document types. Templates can be shared across the team and reused across hundreds of documents.
Firms build their own extraction workflows
Legal teams define their field set, ordering, and required-field rules without writing code. The platform then runs that template across any matching document type.
Stripe-backed subscription and account management
Multiple subscription tiers, team management, invoicing, and plan changes happen in the same workspace lawyers use to review documents.
A bespoke AI pipeline shaped by the lawyers who use it.
We took the engagement from a strategic discovery sprint into a full-scale production application, working closely with the partners doing the actual review work. Three principles shaped every design decision.
Verification first: every extracted field links back to the source paragraph in the PDF, so the user can prove the answer in two clicks. Document intelligence: a custom OCR pipeline plus a bespoke LLM workflow tuned to low-quality scans and handwritten amendments, rather than a wrapper around a generic chat model. Firm-defined extraction: a template engine that lets each firm specify exactly which fields matter for their work, without writing code.
Iterative design with partners and operators ran throughout the build. Early Figma prototypes shaped the split-screen verification view and the template authoring flow. Multi-language support followed the cohort's caseload. Stripe-backed billing and team management let firms onboard themselves once the pilot closed.
- Custom OCR tuned to poor-quality scans and handwritten amendments.
- Bespoke LLM workflow benchmarked against partner-graded leases.
- Verification UI with one-click jump to the source paragraph.
- Template engine for bespoke extraction workflows on any document type.
- Multi-language support across ten-plus languages.
- Stripe-backed billing with team and subscription management.
- ISO 27001 controls and UK data residency end-to-end.
Early prototypes that shaped the verification workspace.
Iterative design started in Figma with senior partners doing the actual lease work. Two early mockups set the shape of what eventually shipped: the question-driven review interface and the split-screen verification view.
How BAiSICS compares to alternatives.
Time to review a commercial property lease, end to end. Tested on the same set of historical leases, marked up by senior partners. Accuracy is the share of extracted fields that match the partner's read.
Minutes per document (lower is better)
92% faster, 5x the capacity, same team.
Processing speed
200 pages of content in around 10 minutes, against the industry's four-hour standard.
Cost efficiency
Operational cost down roughly 60%, with multi-language support across ten-plus languages and zero downtime in production.
User satisfaction
95% satisfaction rate across the pilot cohort. Adoption held through go-live and into production.
Market impact
Now the default tool for the cohort's commercial-lease work. Second AI engagement with the same client already underway.
Having completed our second major AI agent development project with OpenKit, I can confidently say they're the real deal. Our system analyses large, complex and poor quality documents with remarkable accuracy even surpassing leading AI legal tools. The transparency is fantastic; you can instantly verify every AI output against source documents, which builds trust with our users. This isn't just about saving time, it's transformed how we operate as a business. Throughout the project they went above and beyond by proactively improving the platform based on how our users actually work. The UI and UX kept getting better with each update, showing they really understood our customers' needs.
Christopher Wright Director · BAiSICS
BAiSICS has revolutionised our approach, dramatically accelerating the review process for commercial leases and other property documents. We now rely on it for nearly every client instruction. The platform is straightforward, user-friendly, and highly effective. It's an excellent product that excels at what it does.
Working with OpenKit to develop our custom AI agent services has been transformative for our business. What impressed me most was how their AI consulting expertise turned complex AI capabilities into intuitive tools that our team actually enjoys using. Their solution delivers exceptional performance for our specific needs compared to frontier models and competitors while remaining remarkably easy to verify and trust. What sets them apart is their genuine partnership approach - they're not just trying to meet a checklist of needs, but actively improving our product with thoughtful suggestions and innovative features. After two successful projects, we're already planning our third with OpenKit. They've proven invaluable in helping us adapt AI technology across different aspects of our business, and I highly recommend them to any organisation looking to implement practical, powerful AI solutions.
How we delivered it.
Stack
Capabilities
Compliance
From scoping to live.
- Strategic discoveryAudit of existing review process across two firms. Benchmark of generic AI tools against actual lease documents. Weeks 1-2
- Pilot buildCustom OCR plus verification UI. Evaluated against historic leases marked up by senior partners. Weeks 3-8
- Closed betaSix law firms and property consultants running every commercial lease through the platform. Weeks 9-14
- Full-scale productionTemplate engine, billing, multi-org support. Now the default tool for the cohort's commercial lease work. Weeks 15-24
- Ongoing partnershipQuarterly model refresh, new document types added, second AI engagement underway. Ongoing
Read another story
All case studies
Bring your team's next AI project to a 30-minute call.
No deck. We listen, sketch a delivery shape, and tell you honestly whether AI is the right tool for the problem.
