Automated contract data extraction, and where the fields end up
Contracts arrive as they already do. We build the extraction in a business day. It costs nothing until it runs.
Automated contract data extraction reads signed agreements and returns their values as structured fields: the parties, the dates, the term, the renewal and notice periods, and whatever else your process needs. A prebuilt model returns the entities on its published list. A built extraction returns the fields you name, and writes them into the system your team already works in.
Fabrx builds the API. Landing it in a given system is done by partners, so the only requirement is a system that accepts an API call.
What moves, and what it becomes
Master service agreements
The term, the renewal window and the notice period, which are the three values the business needs and three that are written as prose in the body of the agreement.
Dates your renewals calendar can act on
Procurement
Customer order forms and signed quotes
Price, quantity, start date and the signature block, usually arriving as a scan of a printed copy of a document you generated.
Values on the opportunity or the order record
Sales operations
Amendments and addenda
A single page that changes a date set in the agreement it amends. Read on its own, it produces a record that looks complete and holds the superseded date.
A correction applied to the agreement it belongs to
Contract administration
Non-disclosure agreements
High volume, low value each, and still worth the term and the governing law being on a record somewhere when counsel asks which ones are still running.
A searchable register with dates on it
Legal operations
Leases and property agreements
Rent, review dates, break clauses and indexation, often in a document that has been scanned, posted, scanned again and signed in three places.
Obligations on the property record
Finance
Statements of work
Scope, rate, duration and the acceptance terms, written to a different template by every supplier who sends one.
Commitments your project and billing systems can see
Delivery
What happens to one document
Your documents
A contract arrives. Your documents
A shared mailbox, a signing platform, a drive full of PDFs, or a scan posted in by a supplier who still works that way.
T+0
Fabrx
Identify the agreement type. Fabrx
A mixed folder is sorted before anything is read, so an amendment and a master agreement are not treated as the same document.
If An agreement type nobody described
Hold for review. Fabrx. Needs a person.
A named person on your side sees the contract, the clause in question and what the extraction read, before anything is written.
needs a person
Then it rejoins the path above.
Read the fields. Fabrx
Fields are read by what they are, wherever they sit, so a supplier using their own template does not need a rebuild.
If A value the contract cannot settle
Hold for review. Fabrx. Needs a person.
A named person on your side sees the contract, the clause in question and what the extraction read, before anything is written.
needs a person
Then it rejoins the path above.
If An amendment with no parent agreement
Hold for review. Fabrx. Needs a person.
A named person on your side sees the contract, the clause in question and what the extraction read, before anything is written.
needs a person
Then it rejoins the path above.
Send the payload. Fabrx
A structured record, the same shape every time, either written to your system directly or posted to an endpoint you own.
Where we hand over
Two ways, settled on the call. We post the structured record to an HTTPS endpoint you own and your side writes it, through Make or your own middleware, in which case we hold no credentials to your systems. Or we call your system directly, against credentials you authorise and can revoke at any time.
Your CRM or ERP
Fields on the record. Your CRM or ERP
Your objects, your validation rules, your automation. The team works where it already works.
What we do, and what stays yours
Fabrx builds and runs
- Reading the document, whatever shape it arrives in
- Identifying which agreement it is before anything is read out of it
- Pulling the fields you named, and checking what the contract can confirm on its own
- Flagging anything that needs a person, with the document attached
- Building the connection, and rebuilding the extraction when you need a new field or a new agreement type
You keep
- What the terms mean and whether they are acceptable, which is a legal decision and stays with your people
- Telling us which agreement types to expect, which we agree with you during the build
- Which record the values land on, and the permissions the write runs under
- The call that writes the payload, run by you or by your partner
- Somebody to look at the output and tell us whether it is right
When a document is wrong
A pipeline with no failure path is a diagram of a good day. These are the ones that happen.
The scan is unreadable
The document stops before extraction and is never guessed at.
Who sees it. Whoever sent it, by reply, with the page that failed.
- Rescanned and reprocessed
- Keyed in by hand, as it is today
The contract refers to a schedule that was not attached
The fields on the pages you have are returned, and the ones that live in the missing schedule are marked as absent.
Who sees it. Your side, in the exception report, usually the same day.
- Schedule chased and the record completed
- Accepted as it stands
An amendment arrives without the agreement it amends
It is held instead of being written as though it were a new agreement, because applying a changed date to nothing is worse than waiting.
Who sees it. Contract administration, with both documents once the parent turns up.
- Matched to its parent and applied
- Loaded as a standalone record
A field you need is genuinely not stated in this contract
The record arrives with that field empty and marked. Nothing is inferred to fill it.
Who sees it. The person who owns the record in your system.
- Filled from another source
- Chased with the counterparty
Automated contract data extraction, four ways
A person reading each contract, a prebuilt model behind an API, a contract management system with extraction attached. What each one returns, and what it asks of you first.
What it is
Reading each one by hand
A person opens the contract and types the values into a record.
A prebuilt contract model
A trained model behind an API that returns a published set of contract entities.
A CLM with extraction built in
A contract repository with extraction attached to the upload step.
Fabrx
An extraction built for your agreements, landing fields on your records.
Which fields come back
Reading each one by hand
Whatever the person was told to look for, as evenly as a person manages.
A prebuilt contract model
The published list. Parties, jurisdictions, an identifier, a title and line items.
A CLM with extraction built in
The fields the product defines, a fixed list you can see before you buy.
Fabrx
The fields you name on the call, including ones no published list carries.
When the field you need is not on the list
Reading each one by hand
You tell the person to look for that too.
A prebuilt contract model
You train a custom model, which is a project with its own samples and its own timeline.
A CLM with extraction built in
You raise it with the vendor, or map it to a free-text field.
Fabrx
You ask us. Adding a field is a rebuild on our side, and the build is free.
Who configures it
Reading each one by hand
Whoever wrote the instructions.
A prebuilt contract model
Your developers, against the API.
A CLM with extraction built in
Your administrator, inside the product.
Fabrx
We do, from about twenty of your contracts.
Where the values end up
Reading each one by hand
Wherever the person types them.
A prebuilt contract model
A JSON response that your code has to route somewhere.
A CLM with extraction built in
The product record. Moving them elsewhere is its own integration.
Fabrx
The system you already run, by an API call or a webhook your side receives.
What you need before you can start
Reading each one by hand
Somebody with the hours.
A prebuilt contract model
A cloud account, a developer, and reason to believe the published list covers your contracts.
A CLM with extraction built in
A licence, and an implementation before you know it reads yours.
Fabrx
A 30-minute call and about twenty sample contracts.
Someone has run this
They built solutions for any situation encountered along the way. For us, this type of partnership matters.
- 300+
- Field agents using it every working day
- 4,000+
- Documents processed every month
- 4 yrs
- In production, on the same workflow
What a pilot measures
Measured on your documents
Field-level accuracy
Every field on every sample contract, scored against your correction.
Sample size
How many of your real contracts we ran. You choose which.
Exception rate
The share that stopped for a person, and what stopped them.
Build time
Calendar time from your samples reaching us to the pipeline running.
What we ask you for
How long does one contract take today, from arriving to being on the record?
It is the only honest denominator for any claim about time saved.
How many do you process in a month?
It sets the pricing band, because the rate depends on volume.
Which fields do you need that your current tool does not return?
It is usually the reason you started looking, and it decides what we build.
Which record should the values land on, and who owns it?
The mapping and the permissions stay yours, and the build goes faster when the answer exists before the call.
What it costs
- 01
We build it
We build the extraction, usually within one business day of receiving your sample documents.
- 02
You check it
It costs you nothing until it runs on your real documents. If the accuracy is not good enough, nothing goes into production and nothing is charged.
- 03
Then you pay
A prepaid package priced per document, valid for a calendar year. The rate depends on the document type and your monthly volume.
- No subscription
- No per-seat charge
- No setup fee at any volume
Questions about contract data extraction
What is automated contract data extraction?
Software reads a signed agreement and returns its values as structured fields, so the parties, the dates, the term and the money land on a record without anybody retyping them. The work it replaces is a person opening the PDF and copying values across.
What data can be extracted from a contract?
The parties and their signing entities, the effective and expiry dates, the term, the renewal and notice periods, the governing law, the value and the payment terms are the common set. Which of those your process depends on is settled on the call, because that list is yours and it is what we build to.
Can it read scanned contracts?
Yes. A scan, a photograph and a digital PDF all go through the same pipeline. A page that cannot be read stops before extraction and comes back to you with the page that failed attached.
What is the difference between contract data extraction and automated contract review?
Extraction returns what the contract says. Review forms a judgement about it, flagging risk and comparing clauses against a standard. The two are often sold together and they answer different questions. Fabrx does the extraction, and what the terms mean stays a decision for your people.
Do I need a contract management system to extract contract data?
No. A CLM is useful when you want contracts to live somewhere central, and the extraction inside one lands values in that product. If the values are needed in the ERP or CRM you already run, they can be delivered there instead, by an API call or a webhook your side receives.
Does it work on contracts that are not in English?
Yes. A contract in another language is read the same way. Send a few in with your samples and the build covers them alongside the rest.
Related
- automatic document processing
The same work described by its five steps, from capture through to the system it lands in.
- data extraction companies
How the vendors in this market differ, and what to ask each one.
- how the build works
The call, the sample files, and what happens in the business day between them.
Send us twenty contracts
We build the extraction and show you what it got right, field by field, before you pay anything.