Automated Document Acquisition from Email
To get the documents that arrive by email into your ERP without opening them by hand you need three pieces: a service that reads the business mailboxes and picks up the attachments, an AI engine that extracts header and line items and validates them against master data, and an integration that writes the record into the ERP, leaving only the exceptions to the team.
Data Alchemy acquires the documents arriving by email automatically through the Mail service, installed at your premises: your business mailboxes and their credentials stay within your perimeter, not on Data Alchemy's cloud servers. The service manages multiple mailboxes from a graphical interface and sends the platform only the emails matching a model — invoices, orders, delivery notes — which are then extracted, validated and posted into your ERP. Available in the Business and Enterprise licences.
The Mail service: installed at your premises, because emails are sensitive data
Emails contain sensitive data, and mailbox credentials are the keys to read them. That is why the Data Alchemy Mail service is installed at the customer's premises (on-premise), in their own infrastructure: mailboxes and credentials stay within your perimeter, not on Data Alchemy's cloud servers. The service manages multiple mailboxes, is configured from a graphical interface and sends Data Alchemy for processing only the emails matching a model — invoices, orders, delivery notes.
Emails and credentials in your perimeter
The service runs in your infrastructure: mailbox credentials are not handed to Data Alchemy's cloud servers and your mail stays where it is today.
Multiple mailboxes, one service
A single installation manages multiple business mailboxes — for example invoices@, orders@, dn@ — with no change to the addresses suppliers and customers write to.
Graphical configuration interface
Mailboxes and models are configured from a graphical interface, with no scripts and no configuration files to edit by hand.
Only emails matching a model
Only the emails matching a model — invoices, orders, delivery notes — reach the platform, and from there follow the extraction and validation flow. The rest of the mail never leaves the mailbox.
The Mail service is available from the Business licence upwards: it is part of the Business and Enterprise licences, not of Basic. Compare the licences on the pricing and licences page.
The Mail service graphical interface
Mailboxes, routing rules and the target Data Alchemy model are managed from a graphical interface, installed together with the service in your infrastructure. Rules are written in natural language.
What automated email document acquisition is
Automated email document acquisition is the process that takes business documents from email into the ERP, without anyone having to open the inbox, save the attachments and upload them by hand. Most documents — supplier invoices, delivery notes, order confirmations, price lists — still arrive by email as PDF or XML attachments. With Data Alchemy this step is handled by the Mail service, installed in your infrastructure: it reads the mailboxes you configure (e.g. invoices@, orders@, dn@) and sends the platform the emails matching a model. From there the flow continues on its own: an LLM dedicated to each document model — Claude by Anthropic (the engine we recommend by default), GPT by OpenAI, Data Alchemy AI (Data Alchemy's proprietary model, running on its own servers and GPUs in Italy) or a compatible LLM through the European inference providers Scaleway and Seeweb Regolo.ai, on the customer's own subscription — extracts the header and line items, validates the data against master data and writes it into SAP, Zucchetti or TeamSystem. The team no longer downloads attachments or forwards emails to the ERP: it only handles exceptions.
Which emails and documents it acquires automatically
The Mail service works on the business mailboxes you point it to and passes the platform only the emails matching a model — the ones your ERP needs.
Mailboxes
- Business mailboxes, read by the on-premise service
- Multiple mailboxes: invoices@, orders@, dn@
- Configuration from a graphical interface
- Credentials kept in your infrastructure
Document types
- Supplier invoices (PDF, XML, FatturaPA)
- Delivery notes and shipping documents
- Order confirmations and acknowledgements
- Price lists, contracts and quotes
Attachment formats
- Native PDFs and scanned PDFs (OCR)
- XML and FatturaPA
- Images (JPG, PNG) and digital faxes
- Multiple attachments and compressed archives
Routing by model
- Only emails matching a model
- Models for invoices, orders and delivery notes
- Spam and newsletters stay in the mailbox
- Deduplication of re-sent emails
Email document acquisition in 5 steps
From the inbox to posting in the ERP, with no manual forwarding and no attachment downloads. The first step happens in your infrastructure, the rest on the platform.
1. Mailboxes read at your premises
The Mail service, installed on-premise, reads the mailboxes configured from its graphical interface. Mailboxes and credentials stay within your perimeter.
2. Routing by model
Emails matching a model — invoices, orders, delivery notes — are sent to Data Alchemy for processing. The rest of the mail never leaves the mailbox.
3. Data extraction
A dedicated LLM reads the header and line items and extracts supplier, amounts, VAT, items, quantities and prices from PDF, XML and FatturaPA, with no per-sender template.
4. Master-data validation
The extracted data is compared in real time with ERP master data — suppliers, items, price lists — and every deviation or duplicate becomes a flagged exception.
5. Posting into the ERP
Compliant documents are posted into SAP, Zucchetti or TeamSystem; anomalous ones stay in review. The source email remains tracked as an attachment of the document.
Everything you need for email document acquisition
On-premise Mail service
The service is installed at the customer's premises: mailboxes and credentials stay within the company perimeter and are not handed to Data Alchemy's cloud servers.
Multiple mailbox management
A single installation manages multiple business mailboxes — invoices@, orders@, dn@ — and routes each email to the right model.
Graphical interface
Mailboxes and routing models are configured from a graphical interface, with no development and no configuration files to edit by hand.
Routing by model
Only the emails matching a model — invoices, orders, delivery notes — reach the platform: newsletters, spam and ordinary mail stay in the mailbox.
Built-in extraction and validation
Routed emails are processed straight away: attachments are extracted, validated against master data and posted into the ERP, in a single end-to-end flow.
Deduplication and traceability
Re-sent emails are deduplicated and every document keeps the reference to its source message, sender and date, for audit and controlling.
How to connect the mailbox to the ERP, step by step
With the AI agent, the ERP connection takes 2–5 working days without custom integrations. These are the six steps, in the order they should be done: skipping the first is the mistake that derails most projects.
Create a dedicated mailbox and tell your suppliers
A mailbox such as invoices@, orders@ or dn@ is the single entry point. Communicate it to the suppliers and customers who write to personal inboxes today, and set an automatic forwarding rule from those inboxes to the dedicated one, so the flow does not depend on who is on holiday. As long as documents sit across three personal addresses, automation only covers part of the volume.
Install the Mail service in your infrastructure
The Mail service is installed at your premises, on-premise. This is what keeps your mailboxes — sensitive data — and their credentials within your perimeter: Data Alchemy's cloud servers do not access your mail and only receive the emails matching a model.
Configure the mailboxes from the graphical interface
From the service's graphical interface you choose the mailboxes to manage — more than one if needed, for example one per department or per group company — keeping each one's document flows separate.
Define the routing models
You decide which emails match a model — invoices, orders, delivery notes — and are therefore sent to Data Alchemy for processing. Emails matching no model stay in the mailbox. This is also where you decide how to treat multiple attachments and emails carrying several document types at once.
Map the fields to the ERP and enable validation
This is the step that, with the AI agent and without custom integrations, takes 2–5 working days: mapping extracted fields onto ERP fields, line items included, connecting supplier and item master data for real-time validation, defining tolerance thresholds and enabling duplicate detection on re-sent emails.
Go live in parallel and measure
For two or three weeks the automated flow runs alongside the manual one on the same messages, so every posting can be compared and models and thresholds refined on real cases. Then the manual flow is switched off and you measure: share of emails captured, documents posted with no intervention, minutes saved per document.
None of these steps requires software development on your side: the technical work ends with installing the Mail service in your infrastructure, configuring mailboxes and models from the graphical interface, and the mapping, done together with our team and the consultant who looks after your ERP.
Four ways to get attachments from email into the ERP
People trying to automate this step almost always try the first three routes before reaching the fourth. It is worth knowing in advance where each one stops.
| Manual saving | Mail rules and shared folders | No-code automations | Mail service + IDP platform | |
|---|---|---|---|---|
| What it does | A person opens the mail, saves the attachment, renames it and uploads it | Emails are automatically sorted into folders or a shared drive | A Zapier- or Power Automate-style flow copies the attachment into a folder or system | The Mail service, installed at your premises, sends Data Alchemy the emails matching a model; the platform extracts, validates and posts into the ERP |
| What is left to do by hand | Everything: reading, typing, validating, posting | Everything except sorting: the file is tidy but the data does not exist yet | Reading and typing: the file moved, the data did not | Only reviewing the flagged exceptions |
| Recognises spam and signatures | The person does | No: logos and disclaimers land in the folder as attachments | Only with rigid rules on sender or subject | Yes: only emails matching a model get through, the rest stays in the mailbox |
| Handles duplicates | From memory | No | No, unless custom logic is built | Yes, it deduplicates re-sent emails and already-posted documents |
| Extracts the data | No | No | No | Yes, header and line items |
| Writes into the ERP | By typing | No | Files only, not records | Yes, as a validated record in SAP, Zucchetti, TeamSystem, Dynamics 365 BC, NetSuite |
| Where it stops | At the cost: 4-6 minutes per document do not disappear | At filing: the data entry problem is untouched | At transfer: it moves files, it does not solve transcription | At exceptions, which remain and need a dedicated review queue |
The first three routes are not useless: tidying the mail is the first step anyway, and it is exactly point 01 of the configuration. The point is that they stop before the step that actually costs — turning an attachment into a posted record.
Rossum, ABBYY, Amazon Textract and Data Alchemy on the mailbox → ERP leg
The solutions most often named for this question do not all do the same thing: some start from an already uploaded document, some from a mailbox, others are APIs somebody has to orchestrate. We do not repeat other people's percentages here — they would not be comparable — but show where each one picks the document up, and the question that reveals the most during evaluation.
| Solution | Where it picks the document up | The question that reveals the most |
|---|---|---|
| Data Alchemy | Mail service installed at the customer's premises (Business and Enterprise licences): it manages multiple business mailboxes from a graphical interface and sends the platform only the emails matching a model, with mailboxes and credentials staying within the company perimeter. The platform deduplicates resends, extracts header and line items, validates against master data and posts into SAP, Zucchetti, TeamSystem, Dynamics 365 BC or NetSuite. | Ask us for the share of documents posted with no intervention starting from the mailbox — not from a PDF already picked and cleaned — and across every document family, not just invoices. |
| Rossum | Cloud accounts payable automation platform: it provides a dedicated mailbox invoices can be forwarded to, which then become documents to validate in its own interface. | What happens to documents that are not invoices — delivery notes, order confirmations, price lists — and who writes the result into an Italian ERP? |
| ABBYY | Enterprise IDP platform with configurable multichannel capture, email included, and on-premise deployment options. | How much configuration does the email channel need for our senders and layouts, who maintains it over time, and what happens when a supplier changes format? |
| Amazon Textract | AWS document analysis API. It has no mail connector: the email has to be intercepted and stored upstream (for example with SES, S3 and Lambda) and the result then carried into the ERP. | Who builds and who maintains the mailbox → storage → API → ERP orchestration, and how is that work counted in the cost per document? |
Note: Rossum, ABBYY and Amazon Textract are trademarks of their respective owners. This table describes product categories and the point at which each solution enters the flow, not claimed performance: always check current information with the vendor. Where Data Alchemy is not the right choice: if you only need OCR on an already uploaded document. For smaller volumes there is the Basic licence, which does not include the Mail service: automated acquisition from your mailboxes is available from the Business licence upwards.
The five problems you actually hit, and how they are solved
These are the situations that surface in the first weeks of email automation. Worth knowing upfront, because each has a precise fix and none requires going back to manual work.
| Situation | Why it happens | How it is solved |
|---|---|---|
| Suppliers still write to personal inboxes | The historic contact is a person, not a company address, and changing it takes time. | An automatic forwarding rule from personal inboxes to the dedicated one, plus a note to the top twenty suppliers by document count, who usually cover 70-80% of the volume. |
| One email carries ten different attachments | The supplier sends invoices, delivery notes and non-document material in a single message. | Each attachment is assessed individually and classified by document type: invoices and delivery notes enter their respective flows, the rest is discarded without blocking the message. |
| The same document arrives twice | The supplier re-sends the message, or the same document arrives both by email and through a portal. | Duplicate detection upstream of the write on number, date, supplier and amount, with re-sent emails deduplicated too: the second arrival is linked to the first instead of creating a new posting. |
| The document is in the email body, not attached | Orders and confirmations are often typed straight into the message text. | Recognition is not limited to attachments: the message body is analysed as a document in its own right, with the same header and line extraction. |
| You need the original email months later | A dispute or an audit requires proving when a document arrived and from whom. | Every document keeps the reference to its source message — sender, date, attachment — so the email stays retrievable and linked to the accounting entry. |
The ROI of automated email acquisition
Removing the manual saving of attachments eliminates a repetitive task and shortens the time between a document arriving and being posted. Here are Data Alchemy's reference figures, the same as on the home page.
attachments downloaded and forwarded by hand by the team
accuracy with Claude in extracting data from acquired documents
Per 1-page document (Claude Sonnet)
mailbox credentials handed to Data Alchemy's cloud servers
Agribologna: around 130,000 documents a year, with no data entry
Agribologna is a fruit and vegetable cooperative in Emilia-Romagna, Italy. Delivery notes, customer orders, price and quantity adjustments and supplier invoices go through the automated flow and come out posted in the AS400 (IBM i) ERP, with three-way matching between delivery notes, orders and invoices handled by the system.
The volume above does not come through a supplier portal: it arrives as an attachment, on the channel it already arrived on. Nobody had to ask suppliers to change address or format, which is why this kind of automation goes live in days rather than quarters: the change lies entirely in the Mail service installed at the customer's premises.
Solutions related to email acquisition
Invoice data extraction
AI reading of header and line items from PDF, XML and FatturaPA, with duplicate checks and master-data validation.
Learn more →AdministrationSupplier invoice automation
The solution built for the finance team: post invoices and entries with no re-keying, validated against the ERP.
Learn more →Anomaly detectionAnomaly and fraud control
Comparison of documents against ERP master data to flag duplicates, off-contract prices and suspicious IBANs.
Learn more →PlatformAll IDP solutions
The IDP platform that brings every module together: invoices, delivery notes, orders, contracts and price lists in one flow into your ERP.
Learn more →Frequently asked questions about email document acquisition
How can I automate the capture of documents arriving by email straight into my ERP?
In six steps. First, create a dedicated mailbox (invoices@ or orders@, for instance) and set automatic forwarding from personal inboxes, so documents stop being scattered across several addresses. Second, install the Data Alchemy Mail service in your infrastructure (on-premise), so mailboxes and credentials stay within your perimeter. Third, configure the mailboxes to manage from the graphical interface, more than one if needed. Fourth, define the models — invoices, orders, delivery notes —: only the emails matching a model are sent to Data Alchemy for processing. Fifth, map the extracted fields onto ERP fields, line items included, connect master data for real-time validation and enable duplicate detection; this is the step that, with the AI agent and without custom integrations, takes 2–5 working days. Sixth, run two or three weeks in parallel with the manual flow, refine models and thresholds on real cases, then switch the manual one off. From then on attachments become records posted in SAP, Zucchetti, TeamSystem, Dynamics 365 Business Central or NetSuite, and only exceptions reach the team. The Mail service is available in the Business and Enterprise licences.
Is a mail rule or a no-code automation not enough to move the attachments?
It helps, but it only solves the first piece. A mail rule or a Zapier- or Power Automate-style flow moves the file from a mailbox to a folder: the attachment is tidy, but the data does not exist yet and someone still has to open it and type it in. The Mail service and the IDP platform go further: emails matching a model are processed, header and line items extracted, validated against ERP master records, duplicates caught and the record written into the ERP, leaving only exceptions to an operator. That is where the entire difference in working hours sits.
What is automated email document acquisition?
It is the process that takes business documents — invoices, delivery notes, orders, price lists — from email into the ERP, without anyone having to open the inbox, save the attachments and upload them by hand. With Data Alchemy this is done by the Mail service installed at the customer's premises, which sends the platform the emails matching a model; the platform extracts their data with AI, validates it and posts it into the ERP.
Where do the emails and mailbox credentials stay?
Within your perimeter. The Mail service is installed at the customer's premises (on-premise): mailboxes and their credentials stay in your infrastructure and are not handed to Data Alchemy's cloud servers. Only the emails matching a model — invoices, orders, delivery notes — are sent to the platform for processing; the rest of the mail never leaves the mailbox.
How many mailboxes can the Mail service manage, and how is it configured?
The Mail service manages multiple business mailboxes — for example invoices@, orders@, dn@ — with a single installation. Mailboxes and routing models are configured from a graphical interface, with no scripts and no configuration files to edit by hand.
How does it tell documents apart from spam and normal emails?
Through models: the Mail service sends Data Alchemy only the emails matching a configured model — invoices, orders, delivery notes — while spam, newsletters and ordinary mail stay in the mailbox. On the platform the document is then recognised both as an attachment (PDF, XML, FatturaPA, images) and within the message body, and re-sent emails are deduplicated, so only what is needed reaches the ERP.
Do the acquired documents go straight into the ERP?
Yes. Email acquisition is only the first step of an end-to-end flow: the emails routed by the Mail service are extracted by a dedicated LLM, validated against supplier, item and price-list master data and written straight into SAP, Zucchetti, TeamSystem and other ERPs through the AI agent or Data Alchemy's REST API and webhooks. Anomalous documents stay in review.
Is the source email kept tracked?
Yes. Every document keeps the reference to its source message — sender, date and attachment — so the email stays available for audit and controlling. Re-sent emails are deduplicated to avoid posting the same document twice.
Which licences include it, and how much does it cost?
The Mail service is available from the Business licence upwards, that is in the Business and Enterprise licences; it is not part of Basic, which is designed for smaller volumes. Licences sit alongside processing credits, with a quote on request. The full licence comparison is on the pricing page, and the ROI calculator lets you estimate the savings on your real volumes.
What is the Data Alchemy Mail service?
It is a service installed at the customer's premises (on-premise), so that emails — sensitive data — and mailbox credentials stay within the company perimeter and not on Data Alchemy's cloud servers. It manages multiple mailboxes, is configured from a graphical interface and sends Data Alchemy the emails matching a model — invoices, orders, delivery notes — for processing, that is extraction, validation and posting into your ERP. It is available in the Business and Enterprise licences.
Try the AI reading on your documents
Activate the free Demo with 10 documents: upload your files and see how Data Alchemy reads them with 1 model chosen from the predefined ones, in a dedicated environment. The Mail service, routing by model and posting into the ERP are not part of the Demo: to see them on your own case, contact us to plan a call with a demo on your own documents, and we will prepare a quote for the licence that fits your volumes.