Documents that need reading
Invoices, delivery notes, forms and reports read and checked against your rules, with anything unclear flagged.
When off-the-shelf almost works, we build the missing piece.
Sometimes the right answer is a small app, a purpose-built workflow or a carefully scoped AI tool. We design around the job rather than bend your people around a product.
Build or buy?
Off the shelf
- CoversAbout 80% of the job
- Workarounds3 spreadsheets and an inbox
- GapField evidence never reaches the office: Needs a look
Purpose-built
- BuildsOnly the missing piece: Settled
- KeepsThe systems you already run: Settled
- LeavesSafety and sign-off with people: A person decides
AI prepares. People decide.
Custom builds are where AI earns its keep: reading documents, interpreting requests, preparing a recommendation. They’re also where it needs the clearest limits. Every build has rules for what runs on its own, what is prepared for review, and what a person decides.
“We can nearly do this” is a costly place to stay.
- Too many workarounds. Employees fill gaps between systems with email, forms and memory.
- Too big a platform. A full replacement adds training, cost and disruption for one missing process.
- Missing ownership. The clever prototype never becomes something the team can run.
Where it helps
Three jobs for a custom build.
Decisions at volume
Routing, prioritising and checking hundreds of items a week with a decision model, a confidence score and a human fallback.
Work that needs preparing
The history gathered, the context assembled and a draft ready, so a person starts informed instead of searching.
Build the smallest thing that completes the work.
Example: field service workflow
- Capture the realityMobile forms and evidence
- Connect the recordsJob, worker and office information
- Review & approveA personPeople handle safety and exceptions
- Keep it workingDocumentation and ownership
See it run: Tina sketches this workflow, step by step, and shows which parts a person keeps.
What we build
- Workflows in Python with lightweight interfaces, hosted and monitored
- AI for reading, interpreting and preparing, chosen by testing on your own examples
- Decision steps with confidence thresholds: nothing runs on its own below the line you set
- Integrations with the systems you already run
- Review screens where a person approves, edits or rejects
- Logs and evaluations that make the system inspectable
- Testing, operational handover and support options
What stays with your team
- Consequential decisions
- The relationship with the customer
- Anything below the confidence threshold
AI prepares. People decide. Routine actions run only within limits you agree.
Where off-the-shelf stops.
Illustrative architecture — the shape, not a fixed design
What you already run
- Field apps and photographs
- Supplier portals
- Whatever has an API
What carries the work
- Python services for the work no product covers
Power Automate where a supported connector is the better answer
AI Builder document and image reading inside the platform- Observability so a failure is visible, not silent
What stays with people
- What gets promised to a customer
- Quality and safety calls
- Commercial decisions
Microsoft product icons are used under Microsoft’s icon terms for architectural diagrams. Product names belong to their owners.
Usually built with
Python- APIs & webhooks
- AI Builder
- Forms & capture
Technologies we work with. Product names and logos belong to their owners, who do not endorse Boring Future.
The evidence. Each with where it came from.
Boring Future delivery
Purpose-built, not platform-first
Delivered client application; no quantified outcome claimed.
A field-operations app built for the people using it.
- The problem
- SJ Roofing needed to connect the documentation engineers completed on site with the records managers used in the office.
- What was built
- Delivered Roofers Edge, a tailored mobile-first application covering field forms, daily reports, supporting documents, timesheets, photographs and invoice preparation.
- What changed
- The result was a joined-up field-to-office workflow. No numerical saving has been published for this project.
The same project is shown on Workflow & process automation and Trades & field services: one piece of work, not 3 separate cases.
Before we build anything custom
- You probably don’t need an AI agent for this. If rules will do, we use rules.
- A model never takes an action you can’t reverse without a person approving it.
- If Microsoft 365 or a tool you own already covers it, that’s cheaper to run, and we’ll say so.
How it works
- Free conversationFree
- Discovery & Design£2,500 + VAT, fixed
- CompleteFixed scope and fee
- OperateOptional, monthly
Running a business of 5 to 50 people? Essentials starts with a £395 + VAT half-day session
Questions. Anything else, just ask.
Ask us directlyWhich AI models do you use?
The one that does the job best for the cost, tested on your own examples. We benchmark models in the open on our Labs page, including decision models such as TypeSafe’s Jev and OpenAI’s Decisions API.
Where does our data go?
Only where the workflow needs it, with the providers agreed in writing before anything confidential is shared. We start with an anonymised example.
What happens when a model changes?
Models and providers change often. Builds are designed so a model can be swapped and re-tested without rebuilding the workflow.
How do you stop it going wrong?
Rules for what’s certain, review for what isn’t, and a named person for what’s consequential. Failures are logged and routed, not hidden.
Will we own what you build?
Ownership, licences, data handling and handover arrangements are agreed clearly in the written scope.
Do you always recommend AI?
No. We use a simple rule, script or app where that is the safer and better answer.
Can we start small?
Yes. We normally start with one trigger, one team and one outcome before wider expansion.
Where it’s used. Where to go next.
The best software feels like the obvious way the job should have worked.
Start a free 30 to 45 minute conversation. No deck, no obligation.