Unattended Email OCR Pipeline and ERP Integration
The bottleneck of document volume.
Every day the company receives hundreds of delivery notes and orders by email in completely heterogeneous PDF formats. Administrative staff had to download each file, open the Telematel ERP and key in every item by hand. This non-scalable process created a serious bottleneck as the volume of projects and jobs grew.
Reactive ingestion and Serverless Cloud.
Creation of a 100% automated ingestion channel built onAWSinfrastructure. When an email arrives, the system activates on its own, separates the attachments, orchestrates a set of Lambda functions for intelligent extraction (OCR) of the order lines and injects the clean data directly into theERP Telematel.
Measurable Impact
The engineering behind the impact
Intelligent Document Processing
A computer vision and advanced OCR model designed to extract data from documents without fixed templates:
- Table detection: Dynamic segmentation of complex tables (item lines).
- Field mapping: Unambiguous extraction of tax ID, delivery note no., date, item codes, prices and discounts.
Input and Output
- Input: Emails with varied attachments (PDFs, TIFFs).
- Output: Structured data injected directly into the ERP.
Architecture Orchestrated on AWS
AWS SES intercepts the email and sends the attachment to S3. The trigger s3:ObjectCreated wakes up an Orchestrator AWS Lambda that coordinates extraction and loads the information into the Telematel ERP database through a secure, traceable connection.
Want to automate document ingestion into your ERP?
You don't need armies of administrative staff keying in data by hand. We analyze your incoming document flows and build a serverless Cloud infrastructure that can process unlimited delivery notes in a matter of seconds.