---
title: "Document Automation, Workflow Live in 14 Days | JustAutomate"
description: "A document reads itself or builds from CRM data, runs an approval path with deadlines and lands in one place. First flow 14 days after the workshop."
url: "https://justautomate.ai/en/services/document-automation/"
lang: en
---

Process automation

# Document automation without retyping data

A supplier invoice, a contract, a proposal, a meeting record, an HR document. The workflow reads what arrives from outside, or builds the document from data you already hold in the CRM and accounting. Then it moves it through approvals with a deadline and a stand-in, and files it in one place, named right, with the right access. Nobody retypes anything and nobody hunts for the binding version.

 Book a free consultation [See how the workshop works](https://justautomate.ai/en/services/process-workshop/)

- 30 to 60 minuteshow long it takes to put one contract together by hand: template, CRM details, amounts, dates, PDF, email, filing
- 2 dayswhat a team loses every month on 20 contracts, retyping data that already sits in the systems
- 14 daysfrom the workshop to the first working flow for a single type of document

The problem

## How many hours a month does your team spend retyping data into documents?

A supplier invoice arrives by email as a PDF. Someone opens it, retypes the number, the net amount, the VAT and the due date into the accounting system, saves the file on a drive and passes it on for approval. A client contract is already 30 to 60 minutes of work: the template, counterparty details from the CRM, amounts from the proposal, dates, export to PDF, email, waiting for a signature, filing in the right folder. At 20 contracts a month that comes to two full working days spent retyping what the systems already hold.

The second cost is worse than the time. Nobody knows which version of the contract is binding, because three people made three copies of the template. An approval hangs in the inbox of someone on holiday and no one can see it. A salesperson sent a proposal on last year's price list, because the file lived on their own drive. When an auditor asks for a document, you look for it by asking people rather than by opening one place.

Part of this is not a preference, it is the law. [GDPR 2016/679](https://eur-lex.europa.eu/eli/reg/2016/679/oj) in Art. 5(1)(e) requires personal data to be kept no longer than necessary, and in Art. 32 requires control over who can reach it. An employment contract sitting in a folder the whole company can open breaks both. Pulling the other way, the [eIDAS Regulation 910/2014](https://eur-lex.europa.eu/eli/reg/2014/910/oj) states in Art. 25 that a qualified electronic signature has the same legal effect as a handwritten one. The whole flow can run electronically, signature included, and there is no legal reason left to print anything.

- Two working days a month spent retyping data that already sits in the CRM
- Three copies of the same template and nobody knows which one is binding
- An approval stands for weeks because it hangs in one person's inbox
- A document with personal data sits in a folder the whole company can open

Why now

## What changed, so that a document can now be read by machine?

Until recently, reading a document meant OCR plus rules pinned to where a field sits on the page. One invoice from a new supplier with a different layout and the rule stopped working. A language model reads a document by its content rather than by coordinates, so an odd layout, a photo taken on a phone or an attachment in another language no longer breaks the read.

The second change is the signature. DocuSign, Adobe Acrobat Sign and Autenti all expose APIs, so a document goes out for signature on its own and the status comes back into the CRM without anyone asking whether the client has signed yet. Together with Art. 25 of eIDAS, that means the contract loop closes without a printer and without a courier.

The third change is the AI agent. A workflow will police the path, but it will not answer the question how many invoices have been waiting for approval longer than three days. BOND, our company AI agent, answers that from your own company data in 5 to 10 seconds, with nobody logging into three systems to check.

What changes

## What does the team handling documents actually get?

- ### Data out of a document, without retyping
The model reads a PDF, a scan and an email attachment, pulls out the number, the amounts, the parties to a contract and the dates, then writes them where they belong. A person confirms the read instead of performing it.
A supplier invoice moves from an attachment into the accounting form in under a minute
- ### The document builds from data you already hold
A template in Google Docs or Word carries marked fields, and the workflow fills them from the CRM, a spreadsheet or the HR system. What starts the document is an event in a system, not a request on Slack.
The trigger can be a deal closing in the CRM, a project ending or a new hire starting
- ### Approval follows a path, not an inbox
The path knows the value thresholds, the approvers, the deadline and the stand-in for someone on holiday. A document with no decision after the agreed time escalates to a manager instead of lying there quietly.
Every document has a state: waiting on someone, returned with a comment, or approved
- ### One binding version
The template and the price list live in one place, and the document is built from that place every single time. The change history says who edited a clause and when.
A change to the template applies from the next document, with no walking round people's drives
- ### The file lands where it belongs
Once approved, the file goes into the client folder or the year folder, named to a single convention, and into whichever system needs it: CRM, accounting, warehouse. Access follows the role, not who remembered to share it.
One naming convention, access by department, in line with Art. 32 of the GDPR
- ### Questions about a document, in a plain sentence
Is the contract with this client signed. How many invoices have been waiting for approval longer than three days. Which version of the price list went into that proposal. An AI agent answers from company data in 5 to 10 seconds.
An answer in one sentence, without logging into three systems or asking on Slack

And in your company

## What if our documents are different?

Four sentences we hear most often, and what we say back to them.

- We issue a dozen or so documents a month, that is too few to automate.
What counts is not the volume but the time one document takes and how often it gets stuck. If 20 contracts eat two working days and every third one hangs for weeks on approval, the flow pays for itself. If it does not, we say so on the consultation and we stop at one decent template.
- Our contracts are too bespoke for a template.
Almost always it is two or three clauses that are bespoke, not the whole document. The fixed part gets built from data, the variable part stays with a person, with a visible slot for the insert. Documents that genuinely cannot be templated we simply leave alone.
- Our accountant checks every invoice by hand anyway.
And she should, because that is her responsibility. Automation does not take the decision away from her, only the retyping: the data from the PDF is already in the form and she confirms it or corrects it. During the pilot we measure how many reads needed a correction, and only that number decides what later goes through without confirmation.
- We do not want to send scans of contracts to a model.
That is a fair concern and we settle it in the workshop. Documents can be read in an environment that does not train on your data, the fields passed over can be cut down to the ones actually needed, or sensitive documents can stay out of the read entirely while we automate only their routing and filing.

Templates and archive

Google Docs, Google Drive, Microsoft Word, SharePoint

Electronic signature

DocuSign, Adobe Acrobat Sign, Autenti

CRM

HubSpot, Pipedrive, Salesforce, Zoho

Finance and commerce

Xero, sevDesk, Fakturownia, Comarch Optima, BaseLinker, Shopify, PrestaShop

Approvals and team work

Slack, Microsoft 365, Google Workspace, Notion, ClickUp, Jira, Asana

Data and the automation layer

BigQuery, Google Sheets, SQL databases, n8n, Make, Zapier, custom API integrations

## Documents that most often go into the first workflow

The three columns answer the three questions that have to be settled for every document: where the data comes from, who takes the decision, and where the file has to end up. The order of rollout is agreed in the workshop, by how many of each you produce in a month.

| Document         | Where the data comes from                                                                     | Where it lands after approval                                              |
| ---------------- | --------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------- |
| Supplier invoice | read from a PDF or a scan in an email: number, amounts, VAT, due date                         | the accounting system, plus the year folder on the drive                   |
| Client contract  | template plus counterparty details from the CRM, VAT number checked against the VIES register | signature in DocuSign or Autenti, the client folder on the drive           |
| Sales proposal   | line items and prices from the CRM, terms from a single current price list                    | PDF to the client by email, attached to the deal in the CRM                |
| Meeting record   | transcript from Tactiq or Fireflies, action points taken from the decisions                   | Notion or Google Drive, tasks in ClickUp or Jira                           |
| HR document      | data from the HR system, template of the amendment or the certificate                         | the employee file, open to the HR team only                                |
| Client report    | BigQuery, GA4, a sheet with campaign or project results                                       | PDF by email on an agreed schedule, a copy in the client folder            |
| Customer order   | read from an email attachment or a file sent by the counterparty                              | line items in BaseLinker or the warehouse system, file kept with the order |

A first workflow usually covers one or two types of document, the ones with the highest count per month. The rest goes in once the first one runs and has a measured result.

How we work

## Five steps from a pile of files to one workflow

Process first, tools second. Every step ends with something you can judge before you decide on the next one.

1. ### Workshop
We take today's flow apart: which documents get produced, how many a month, who assembles them, who approves them, where they sit and how long they have to sit there. We time a single document, so that after the rollout there is something to compare against.
2. ### Flow design
Which documents go into the first flow, templates with marked fields, an approval path with value thresholds and stand-ins, a naming and folder convention, a retention period. We decide which fields the model may read and which have to come from a system, because guessing them is not allowed.
3. ### Pilot on real documents
The workflow runs alongside the current way of working, with nothing switched off. We push the last few weeks of documents through it, including the awkward ones: a phone photo of a scan, an invoice with an unusual layout, a contract with an amendment, a proposal in a foreign currency.
4. ### Production and training
Once the results check out we switch to production with error monitoring. The person who owns documents learns to change the template, the approval threshold and the list of approvers without us. Updating a price list stops being a ticket in our queue.
5. ### Measurement at 30 and 90 days
We come back with numbers: how many documents went through the flow, how many got stuck and where, how many reads needed a human correction. Then we tune the path to what actually turns up, not to what we assumed in the workshop.

Before and after

## How the week changes for the person who owns documents

### Before

- A salesperson assembles a contract by hand, 30 to 60 minutes each
- A supplier invoice waits for approval in the inbox of someone on holiday
- Nobody knows which copy of the template is the binding one
- At an audit you search drives and inboxes for the document

### After

- The contract is built from CRM data the moment the deal changes stage
- A document with no decision after the agreed time escalates to a manager
- There is one template, and a change to it applies from the next document
- The file sits in the client folder, named to a convention, open by department

The first flow runs 14 days after the workshop. Document time and the number of stalls get compared at 30 and at 90 days, against the measurement taken in the workshop.

For the person signing this off

## What an operations director gets out of it

The questions where is that contract and who approved it stop landing on your desk, because the system answers them rather than someone's memory. Document turnaround becomes a number you can put on a slide, instead of an impression picked up in a corridor.

The second thing counts at an audit and at an inspection. You know where every document sits, who has seen it and how long it stays. That is an argument you do not have today, and it takes personal risk off you when somebody asks about an employee file or a contract from three years back.

- The approval path written down and versioned, independent of any one person
- Access to documents by role, in line with Art. 32 of the GDPR
- A pilot alongside the current way of working, so the old flow runs until testing ends
- You change the template and the approval threshold yourself, without a ticket to us

What it costs

## The workshop has a published price, the rollout is quoted after it

The process workshop costs **EUR 3,500 net per week** and runs from one to seven weeks. With documents it usually comes to one or two weeks, because a document flow touches several systems at once and the approval path has to be written down together with its exceptions. We quote the rollout after the workshop, once it is clear how many document types go into the first flow and what state the templates are in.

Tool costs you pay directly to the provider: [n8n](https://docs.n8n.io/), Make, the e-signature service, and the model that reads the documents, where a single query costs the provider roughly EUR 0.005 to 0.07\. No margin from us on any of it. With documents that cost depends mostly on the number of signatures and pages per month, so we give you the order of magnitude in the workshop rather than on the first invoice.

- ### Process workshop
EUR 3,500 net
per week, and with documents that is usually one to two weeks
A self-contained piece of work. It ends with a map of the flow and an approval path you could roll out without us.

  - An inventory of documents with monthly counts and the time one copy takes
  - An approval path with value thresholds, deadlines and stand-ins
  - A naming, folder and retention convention
  - A list of automations in payback order, each with a recommended tool
 Start with the workshop
- Most popular
### Workshop and document workflow
Quoted after the workshop
scope follows from the number of document types and systems involved
The option most companies choose. A working flow 2 to 4 weeks after the workshop, tested on your own documents from the last few weeks.

  - Everything in the workshop
  - Reading data from PDFs, scans and attachments, with confirmation on the critical fields
  - Templates with variable fields, plus document generation from system data
  - Approval path, stand-ins, escalation and writing back into the target systems
  - Pilot alongside the current way of working, switchover with monitoring, team training
 Talk about a rollout
- ### Document workflow and an AI agent
Quoted after the workshop
for companies that ask about document status every day
Everything in the rollout plus BOND, the company AI agent that answers questions about documents in a plain sentence.

  - Everything in the document workflow
  - An AI agent wired into the archive, the CRM and accounting, answering in 5 to 10 seconds
  - Alerts on documents sitting in approval longer than an agreed threshold
  - Agent rollout takes around 2 months, running alongside the working flow
 Ask about the agent

The workshop ends with a document that stands up without us: a map of the flow, the approval path and a list of automations in payback order. If it turns out that one shared template and a tidy folder structure are enough, we say so and show you how to do it. The engagement then ends at the workshop, not at a rollout you do not need.

## Frequently asked questions

What is document automation in a company?

It is a flow in which a document either builds itself from data you already hold in your systems, or gets read by machine when it arrives from outside. From there it follows an approval path with a deadline and a stand-in, and once approved it lands in one place, under the right name and with the right access. People still decide and approve, they just stop retyping data.

How long does a document workflow rollout take?

The workshop is one to two weeks, the rollout 2 to 4 weeks after it. The first working flow for a single type of document is usually visible 14 days after the workshop. It takes longer only when the templates or the folder structure have to be tidied up first, because automation built on a mess simply repeats the mess faster.

Can you read data from an invoice that arrives as a scan?

Yes. The model reads scans and PDFs without a fixed field layout, so an unusual design no longer breaks the read the way position based rules used to break. Amounts and the bank account number we leave for a person to confirm. During the pilot we measure how many reads needed a correction, and that result decides what later goes through without confirmation.

Who approves a document when the responsible person is on holiday?

The stand-in agreed in the workshop. An absence marked in Google Calendar or Outlook reroutes the task automatically, and a document with no decision after the agreed time escalates upward. The flow therefore does not stop at one inbox, and you can see whose desk each document is sitting on.

Is an electronic signature enough instead of paper?

For most commercial contracts, yes. The eIDAS Regulation 910/2014 gives a qualified electronic signature the same legal effect as a handwritten one in Art. 25, and DocuSign, Adobe Acrobat Sign and Autenti all expose APIs, so the signature status comes back to the CRM by itself. Cases where the law demands a special form, a notarial deed for example, we catch in the workshop.

What happens when we change a template or a price list?

You change it in one place and from the next document the new version applies. There are no copies of the template on salespeople's drives, so a proposal on last year's price list stops being possible. The change history shows who edited a clause and when, which helps when a client asks about terms agreed a year ago.

Where do documents with personal data go?

Into a folder with access by role, for example an employee file open to the HR team only. The retention period is agreed in the workshop, because Art. 5(1)(e) of the GDPR requires data to be kept no longer than necessary, while accounting and employment rules set the lower bounds. For reading, we pass over only the fields needed for the decision.

How is this different from contract generation or document archiving?

Contract generation, invoicing from the CRM and archiving are single scenarios, each with its own page and its own quote. This service sits a level above: it covers the whole flow, meaning reading data out of the document, the approval path, template versions and where the file has to land in each of your systems. We usually start from one scenario and add the rest.

## Let us find which document costs you the most

The free consultation takes 30 minutes. You tell us which documents you produce most often and roughly how many a month, from memory is fine. We tell you which one can go into a flow straight away and which one needs one shared template and a tidy folder structure first. You do not have to share systems or send us any contracts at this stage.

We run at most two rollouts at a time, so that someone is genuinely sitting on each of them.

[Book a free consultation](https://justautomate.ai/en/contact/)

**PS.** If somebody in your company asked this week where the signed version of a contract is, that is exactly this problem. Two working days a month of retyping and a week of waiting on one approval is a cost you can see on a single document. That is where the workshop starts. You write directly to Pawel, no reception and no tickets.

## Related

- [Automatic contract generation](https://justautomate.ai/en/automations/automatic-contract-generation/)
- [Automatic invoicing from CRM](https://justautomate.ai/en/automations/automatic-invoicing-from-crm/)
- [Automatic document archiving](https://justautomate.ai/en/automations/automatic-document-archiving/)
- [Invoice automation](https://justautomate.ai/en/services/invoice-automation/)
- [BOND, business AI agent](https://justautomate.ai/en/services/bond-business-ai-agent/)
- [Process workshop](https://justautomate.ai/en/services/process-workshop/)

---

**JustAutomate** · contact: contact@justautomate.ai · form: https://justautomate.ai/en/contact/