Can it write into our accounting or ERP system?

Usually, through that system’s API or an import. We confirm it during scoping, and nothing is written without the approval step you choose.

Can it read scanned or photographed documents?

Often, yes, though accuracy depends on their quality. Testing on your real samples shows how well it reads your documents.

Do we still need people?

Yes. A person approves results before they are recorded, and handles the documents the system flags as unclear.

Which documents can it handle?

Invoices, purchase orders, delivery notes, application forms and reports are common. We confirm which of yours suit it during scoping.

AI document processing

AI document processing reads the invoices, forms and reports someone types into a system by hand today, extracts the fields you need, checks them and sends them on, with a person approving the results before anything is recorded.

We start with a set of your real documents and the correct values for each one. The processing is tested on them, so you see how accurate it is on your documents before you decide, rather than taking a vendor’s word for it. Documents it cannot read clearly go to a person instead of being guessed.

Start a project See an example

Document processing

Illustration of document processing work, with sample content

For example: supplier invoices read from email attachments, checked, and recorded in your accounting system once a person approves them.

An illustration of the kind of work we do, not a client project.

Supplier invoices today

Before

Invoice PDFOpened by hand

Every fieldTyped into accounting

Typing mistakesFound at month end

Read and checked

After

Supplier

Read from the invoice

Invoice date

Read from the invoice

Total

Checked against the order

StatusApprove to record

What document processing includes

Sample set
Invoice from a packaging supplierCorrect values noted
Invoice from a courierCorrect values noted
Delivery note, scannedCorrect values noted
Sample documents with correct values
Real invoices or forms from your business, each with the values a person would enter.
INVOICEA packaging supplierBoxes, 200Tape, 40 rollsTotal due
SupplierA packaging supplierInvoice date2 OctoberInvoice no.From the headerTotalMatches the order
Reading and extraction
Supplier, dates, totals, line items or form fields read from PDFs, scans and photos.
Check before recording1 to review
SupplierA packaging supplier
TotalMatches the order
Tax numberUnclear: please check
Approve and record
Review before recording
A person checks and approves the extracted values before they reach your systems.
Sent after approval
Accounting systemBill created
Google SheetRow added
Your ERPOptional
Output where you need it
Sent to your accounting system, ERP, spreadsheet or database.
AutomationPaused
Manual steps, written down1. Open the inbox 2. Sort by type 3. Reply from the template
A manual route for hard cases
Unclear or unusual documents go to a person, with the reason flagged.

Want this for your business?

Tell us what you need. A founder reads every enquiry and replies with questions or a proposed scope.

Start a project

Who it’s for

Finance teams entering invoices by hand
Supplier invoices typed into accounting, field by field.
Operations teams processing forms
Applications, claims or delivery notes checked and entered by hand.
Businesses receiving many formats
Reports and documents from partners, each laid out differently.

What you get at handover

Everything is handed over in your name, so you can run it, change it or move it to another team.

Handover record Checked by a reviewer

The processing, running in your own accounts
The test documents and the accuracy results
The review screen, and who approves results
The manual route for documents it cannot read
The expected monthly running costs

Illustration of a handover record

What we need from you

A project moves fastest when these are ready. We help you put together anything you do not have yet.

1A set of real documents, with the correct values for each
2Access to the system the results go into
3Agreement on which AI tools may see the documents
4A person to review and approve results

How an engagement runs

Scope document
Scope: customer portal
What we deliver
Customer sign-in, with roles
Order status page
Invoice downloads
Not includedMobile appDone whenTested with your real orders
Agreed by you before the build starts
This week’s build
This week’s increment
Order status page
Ready to view
feat: order status page
test: status for each order type
docs: how statuses are calculated
Next increment: invoice downloads
Review checkpoint
Our reviewer
Checked against the scope
Tests read and run
Ready for you
Your approval
Approve milestoneRequest changes
Handover pack
Code repository
In your account
Credentials record
Transferred to you
Documentation
What was built and how
Runbook
Deploy, roll back, fix common problems
Maintenance available on a monthly plan
Offered as

Scope document

Scope: customer portal

What we deliver

Customer sign-in, with roles
Order status page
Invoice downloads

Not includedMobile appDone whenTested with your real orders

Agreed by you before the build starts

Illustration of what you receive

What affects the cost

We quote after a scoping conversation, so there is no fixed price on the site. These are the factors that move the cost of this work up or down; the proposal states the scope, what is not included and the price.

Technologies we commonly use

AI models for reading text and layout
PDF, scan and photo input
Accounting and ERP connections
Google Sheets
Python
Email inbox connections

Number of document types: Invoices alone is a smaller job than invoices, delivery notes and claims.
Variety of layouts: Many suppliers with different layouts need more testing.
Quality of the documents: Clear PDFs are easier than phone photos or handwriting.
Where the results go: Writing into an ERP or accounting system adds integration work.
Volume: How many documents arrive, which drives running costs.

How we keep AI reliable

Every AI service follows the same five rules, on top of the review every project gets.

Checklist for every AI projectBefore handover

Real examples first
We collect real examples from your business, with the correct answers, before we build.
Tested against them
We test against those examples and show you the results.
A person approves
Anything that sends messages, moves money or changes records has a person approving it, unless you decide otherwise.
Running costs stated
You know the monthly running costs before handover.
A manual fallback
Every automation has a documented manual fallback.

Questions

That depends on your documents, so we measure it rather than promise it: we test on your real samples against the correct answers and share the results before you decide.

Related services

All services

AI automation

Workflows between your tools that remove repetitive manual work.

APIs and integrations

Connect your CRM, ERP, payments, WhatsApp and spreadsheets so data moves on its own.

Internal tools

Replace spreadsheets with tools for tracking, reporting and approvals.

Tell us what you want to build.

Describe the work. We will scope it with you, or tell you plainly if it is not work we should take on.

Start a project All services