Security inside the build
Client-approved tools and infrastructure supported PPL’s working proof of concept.
Read the PPL methodInsurance document processing with AI
Insurance document processing with AI: a working pilot in weeks
Tenhaw turns insurance PDFs into structured, enriched data, with review by confidence and consequence. Test one workflow through a £20,000–£55,000 proof of concept.
Security inside the build. Client-approved tools and infrastructure supported PPL’s working proof of concept. Read the PPL method.
Book a 30-minute call
See the evidence and approach30 minutes with James Rooney to discuss your team, workflow and first scope.
The job to be done
What we can help you deliver
For insurance operations and technology teams testing how contracts and PDFs could become structured information that reviewers can check and use.
Turn document content into structured information
At PPL, the working pipeline extracted, validated and enriched PDF data for an existing business-intelligence dashboard. Test the documents and fields that matter in your workflow.
Give reviewers the uncertain decisions
Route review by confidence and consequence, retaining the source needed to check a field. Agree who can resolve a doubtful value and when processing must pause.
Make a production decision with evidence
A pilot tests the workflow against your documents. Its result informs a separate production scope, including throughput, identity, error handling and operational ownership.
Use a 30-minute call to discuss the outcome your team needs.
Book a 30-minute callWorking together
Build the capability around real work
- 01
Bring real documents and acceptance examples
Choose a bounded document workflow, its data owners and the difficult cases your team already handles. Agree what counts as a useful result before the build.
- 02
Build the pipeline with your engineers
Work inside your repositories and infrastructure. The published PPL pipeline extracts to markdown, narrows to key fields, normalises, enriches against third-party APIs, adds semantic context and thematic grouping, applies business logic and lands in the dashboard.
- 03
Review the result before expanding
Record which examples pass, fail or need a person. The proof of concept ends with a go/no-go decision and a fixed quote for the 90-day outcome programme.
Ask how this approach would fit your data, systems and team on a 30-minute call.
Book a 30-minute callEvidence and approach
Read the evidence behind the approach
Paired engineering at PPL
At PPL, a six-month Embedded Agentic Lead engagement included a two-week greenfield working proof of concept, paired with a client engineer. It did not enter production.
Read the PPL proof of conceptBook 30 minutes to ask James about the work and how it relates to your situation.
Book a 30-minute callScope and price
Choose a starting engagement
Agree the workflow, deliverables and team before work starts. These are the existing engagements this use case can sit within; prices exclude VAT.
Agentic Proof of Concept
Fixed price · £20k–£55k
2–4 weeks
Build and test one AI workflow in your own tenancy.
See scope, team and terms90-day outcome programme
Fixed price · £120k or £180k
90 days
One workflow in production and in use in 90 days.
See scope, team and termsUse a 30-minute call to decide which scope is a sensible starting point.
Book a 30-minute callRead further
Explore the method and related use cases
- Tenhaw, London AI consultancy
- Compare the published engagements
- Follow the document-to-BI approach
- AI in financial services
- Document intelligence questions
- AI agents and workflow automation
- Enterprise AI assistants
- AI product development
- AI training and adoption
- Forward-deployed AI engineers
- AI governance, risk and controls
- AI transformation programme
- AI-native delivery
book a call
Talk through the work you want to change.
A 30-minute discovery call with James Rooney. Bring a workflow, product or team challenge and we will discuss a sensible first scope.
most start with a fixed-price AI Readiness Audit · £44,000 · 4 weeks · working prototypes
// pick a slot · cal.com/tenhaw/professional-servicesLIVE CALENDAR
No time that suits? Email hello@tenhaw.com. Calendar not loading? Open it on cal.com.
Describe the work
Describe the work in a paragraph. James replies with a rough scope and budget range within two working days.
Frequently asked questions
What documents can it handle?
The published PPL proof of concept processed contracts and other PDFs into structured, enriched information for business intelligence. Your own document types and difficult examples need testing before a scope or accuracy claim can be agreed.
How is accuracy controlled?
Use validation and confidence-based routing with the source material available for checking. Test against agreed examples and record failures and cases needing human review. Confidence alone does not establish that a field is correct.
Who reviews low-confidence fields?
The people with the expertise and authority to resolve them. Agree that review role during workflow design, including which consequences require a person even when model confidence is high.
Does it work on our infrastructure?
Yes. Tenhaw builds in your environment, with your data and repositories, under your policies and access controls. Only AI tools approved by you in writing may touch client data.
What does a pilot cost?
Tenhaw’s Agentic Proof of Concept costs £20,000–£55,000 fixed over two to four weeks, excluding VAT. One or two engineers pair with your people on one agreed workflow. Complexity and the state of the data determine the scoped fee.
Is it production-ready after the pilot?
The pilot ends with a working proof of concept and a separate production decision. The PPL example was a two-week greenfield build that did not go into production. Tenhaw’s 90-day programme separately covers one workflow into production and use.