01Agents & automation · 11
Documents read into the ledger, with every field accounted for.
Invoices, contracts, delivery notes, forms and scans, read by models that understand the page, checked against your master data, and posted where they belong, with the exceptions handed to a person and nothing lost on the way.
For finance, procurement, operations and legal teams who key documents into systems by hand, and for anyone whose month-end is a pile of PDFs.
01Overview
ExtractionA document is a layout, not a string. We use models that read tables, stamps, handwriting and multi-page structure, validate what they read against the data you already hold, and attach a confidence that means something to the field it describes.
01/ 02
Extraction
Read the page the way a clerk does, only every time.
A document is a layout, not a string. We use models that read tables, stamps, handwriting and multi-page structure, validate what they read against the data you already hold, and attach a confidence that means something to the field it describes.
- 01
Invoices, contracts, forms and scans
Supplier invoices in a hundred layouts, contracts with clauses to find, application forms with handwriting, and scans of all of them at any quality. One pipeline handles the range, with a document type routed to the extraction it needs.
- 02
Models that read the layout
Line items in tables, totals in the corner, terms on page four and the stamp over the signature are found by models trained on layout, not by a template that breaks when a supplier changes their form.
- 03
Validation against master data
A supplier name is matched to a supplier record, a part number to the catalogue, a tax rate to the schedule and a total to its line items. A field that fails validation is flagged, never quietly accepted.
- 04
Confidence that means something
Every extracted field carries a calibrated confidence, and thresholds are set per field with you: a supplier code is held for review at a stricter level than a description. Low confidence is a routing decision, not a number in a log.
02/ 02
Posting
Into the ledger, with the exceptions where people can see them.
Extraction is half the job. The other half is getting the data into the system of record correctly, matching it against what was ordered and received, and leaving a trail that an auditor can follow from the posted value back to the pixel it came from.
- 05
Exception queues, not silent failures
A document that cannot be read, validated or matched goes to a queue with the page, the fields and the reason side by side. A person resolves it in seconds, and the resolution teaches the next run.
- 06
Posted into the SoloOne ledger
Invoices, receipts and credit notes post straight into the SoloOne purchase and sales ledgers, or into the finance system you run, with the document attached to the entry.
- 07
Three-way matching across purchase, receipt and invoice
The invoice is matched to the purchase order and the goods receipt on quantity, price and terms. A match posts; a mismatch goes to the queue with the three documents open together.
- 08
An audit trail for every field
Every posted value records where on the page it was read, which model read it, its confidence, who reviewed it and what they changed. An auditor can follow any number back to its source.
At a glance
Every capability, at a glance
8 figures, one per capability. Open any to read it in full.
How it runs
From first call to running unattended
Every engagement runs the same five steps, whatever the service.
- 01
Map the process
We sit with the people who do the work today and write down every step, exception and hand-off before anything is built.
- 02
Build against a golden set
A held-out set of real cases, agreed with you, is the bar each build has to clear before it goes anywhere near production.
- 03
Run beside the team
The system runs in parallel with the team for as long as it takes, and every disagreement between them is reviewed together.
- 04
Hand over the keys
The code, the prompts, the evaluation set and the runbooks are handed over in your accounts, under your keys.
- 05
Keep it running
We watch the runs, retrain and repair as the inputs drift, or train your own team to do the same.
Outcomes
What this looks like when it lands.
01 Invoices posted the day they arrive
Supplier invoices read, matched and posted without a keying step, with only the mismatches waiting for a person.
02 A month-end without a pile
Documents flow into the ledger through the month, so closing is a review of exceptions rather than a week of data entry.
03 An auditor who can follow every number
Each posted value traces back to the page, the model, the confidence and the reviewer.
Details
On the spec sheet.
- Service
- Document Intelligence
- Group
- Agents & automation
- Capabilities
- 8
- Engagement
- Map, build, run beside the team, hand over, keep running
- Ownership
- Your accounts, your keys, your region
- Case studies
- Supplier catalogue onboarding and enrichment, Supplier invoices, inbox to posted ledger entry, Customer purchase orders into ERP sales orders, Contractor timesheets and compliance files, Commercial lease abstraction and obligations, Cashless pre-authorisation packets, Motor claims first notice of loss and triage, Periodic KYC re-verification, Admissions document processing
Questions we are asked about document intelligence
Asked before signing.
Do we have to build a template for every supplier?
No. The models read layout rather than positions, so a new supplier or a changed form is handled without a template. Unusual documents go to the exception queue the first time and are learned from there.
What about handwriting and poor scans?
Both are read, with confidence reflecting the quality. Fields below the threshold you set go to review rather than being guessed.
Which systems does it post into?
The SoloOne ledgers natively, and the finance or ERP system you already run through its API. The document is attached to the entry in either case.
How is accuracy measured?
Against a golden set of your own documents, field by field, before launch and on every model change. You see the numbers per field and per document type, not one headline figure.
Where are the documents stored?
In your storage, in your region, for the retention you set. The models run in your cloud accounts or ours under a data-processing agreement, as you prefer.
Start