ERP and CRM
From the page to your system, without retyping
Invoices, receipts and business cards are read here and come out as a file with fixed columns that your ERP or CRM imports. Or through the API, straight into your automation.
01Three ways
Choose by volume, not by system
01 · No setup
An import file
In the workspace, after the read: an ERP file with one row per line item, a CRM file with one row per contact, a vCard for your contacts and Excel for everything else. Map the columns once in your system's import.
Open a document02 · For a steady flow
API and automation
Send a document to POST /api/v1/read and get JSON with the same keys every time. Any automation tool with an HTTP step can call it, such as Make, n8n, Zapier or Power Automate. The start is free, with 100 pages.
Get an API key03 · For large volumes
Fitted to your system
Priority, SAP Business One or any system with its own import template: send us the template and we will look at producing the data in exactly the shape it expects.
Talk to us
02The ERP file
One row per line item, the document's details on every row
The file comes out with fixed English columns, in the same order for every document. A document without a line table comes out as one row. If your system imports headers and lines as two separate files, write to us.
03The CRM file
One row per contact, every detail in its own column
A business card, or any page with contact details, comes out as a row. In a PDF of several cards, one card per page, every card is a row.
And a vCard: the same details in a file that opens in your phone's contacts, in Google Contacts and in Outlook.
04Where the files go
A file import, in the systems you already use
We don't have a ready-made connector for every system. This is how the files get into common ones, through their own import tools:
ERP and accounting
- SAP Business One
- Import through the Data Transfer Workbench, which works with CSV templates. Map the columns to the document's template.
- Priority
- Load interfaces take data from text files. Send us your interface's layout and we will check the fit.
- Hashavshevet and other systems
- Send us the layout of the file you import today, and we will look at producing the data in that shape.
- Excel and Google Sheets
- The file opens as it is, and every document adds rows under the same columns.
CRM and contacts
- Salesforce
- The Data Import Wizard imports contacts and leads from a CSV file, mapping columns to fields.
- HubSpot
- Import contacts from a CSV file, mapping columns to properties.
- monday CRM and Zoho CRM
- Import from an Excel or CSV file, mapping the columns.
- Phone, Google Contacts and Outlook
- A vCard file opens in your contacts.
05API and automation
A steady flow of documents, hands off
- 01Get an API key. The start is free, with 100 pages.
- 02In your automation tool, add an HTTP step: a POST to /api/v1/read, with the key in the Authorization header and the document as a file or a link.
- 03The answer is JSON: fields holds the values under fixed keys such as vendor_name, total and issue_date; tables holds the line items.
- 04Map the fields to an action in your system: a vendor bill in the ERP, a contact or lead in the CRM, or a row in a sheet.
- 05A value that arrives with "uncertain": true can stop for a person to check before it enters the system.
curl https://ocr-ai.com/api/v1/read \
-H "Authorization: Bearer ocrai_sk_…" \
-F "file=@invoice.pdf" \
-F "pages=1-3"06Connect
Tell us about your system
Which system you work with, how many documents a month and what needs to happen with them. We will reply by email.
07Questions
Questions about connecting your systems
Is there a ready-made connector for SAP, Priority or Salesforce?
Not a ready-made one. There are import files with fixed columns, and an API that any automation tool can call. For a connection built around your system, write to us.
What about an invoice with several pages?
Every page of the document goes into the same file: the document's details on every row, and the line items from all pages in sequence.
What happens when a value was not read with certainty?
In the workspace it is marked for checking. In the API it arrives with "uncertain": true, so your flow can stop it for a check.
Do you keep the documents?
No. The file is read in memory and not stored; we keep usage data without content.
What does it cost?
The same page packs as on the site, shared by the browser and the API. For a steady volume of thousands of pages a month, write to us.