Data Alchemy — Software IDP con AI
Industry comparison

Which IDP tools suit Italian manufacturing companies best

For an Italian manufacturer, the right IDP tool is the one that handles the full production document cycle — customer orders, supplier order confirmations, raw material delivery notes, certificates of conformity, supplier invoices — and writes the data straight into the ERP or MES already in use. On that criterion the options split into two families. Data Alchemy and Rossum are full IDP platforms that return validated data; ABBYY is the long-standing choice for structured enterprise projects, on-premise included; Google Document AI and Azure AI Document Intelligence are cloud provider extraction services — outstanding at recognition, but with no knowledge of your ERP, your warehouse or Italian e-invoicing. The practical difference for a manufacturing SME is where the data ends up: in a JSON someone still has to carry into Zucchetti or TeamSystem, or in a goods receipt already posted and linked to the batch. Below are the industry-specific criteria, the assessment of all five tools and the comparison table.

The documents

The six documents that decide whether an IDP works on a factory floor

Assessing an IDP on supplier invoices alone is the most common mistake in manufacturing: it is the easiest document and the least critical. These six are the ones to test it on.

DocumentWhat it must extract and validateWhat happens when it gets it wrong
Customer ordersCustomer item codes to translate into yours, quantities with correct units of measure, requested delivery dates including per line.A misread order becomes a wrong production job: you find out on the line or at shipping.
Supplier order confirmationsAutomatic comparison with the purchase order issued: items, quantities, confirmed prices and promised dates, with variances highlighted.An unrecorded confirmation is a line that stops because the material does not arrive when expected.
Raw material delivery notesSupplier, batch, items, quantities and units of measure, matched against open purchase orders with stock updated.A goods receipt posted late or with the wrong batch breaks traceability downstream.
Certificates of conformity and qualityCertificate number, issuing body, batch, material, test values and tolerances, linked to the inbound batch.Without an automatic link to the batch, the certificate exists but cannot be found when it actually matters: during an audit or a dispute.
Supplier price lists in PDF and ExcelCodes, prices, volume breaks and discount bands extracted in a structured way and loaded into master data.A price list updated by hand once a year means buying at the wrong terms for months.
Supplier invoicesHeader and line items, three-way matching against order and delivery note, duplicate detection and master data validation.It is the simplest document: if a vendor showcases it as their flagship case, that is the signal to ask for the other five.
Criteria

Seven criteria specific to Italian manufacturing

These are the criteria that separate a tool suited to a factory from a tool suited to an office. They are worth more than any generic score.

1. Writing into the ERP or MES in use

The data has to land inside the system you already run — TeamSystem, Zucchetti, SAP, Dynamics 365 Business Central, NetSuite — as a complete record. If the tool stops at JSON, the last mile is a separate project, and it is almost always the longest part.

2. Line items, not just headers

A raw material delivery note or a customer order matters because of its lines. Ask for accuracy split between header and lines: the overall average hides exactly the problem you care about.

3. No templates to maintain

A manufacturer's supplier base changes constantly and every supplier has its own layout. A system requiring a template per format creates a maintenance cost that grows with the business.

4. Batch handling and traceability

In manufacturing a document is not an end in itself: the delivery note hooks the batch, the batch hooks the certificate. Check that the tool can link documents to each other, not just read them one at a time.

5. Italian document and tax context

FatturaPA XML, SDI recipient code, VAT numbers and tax codes, VAT rates and typical delivery note formats. These are things an international tool does not know and that have to be built, with the cost and timeline that follow.

6. Predictable cost on variable volumes

Production has seasonality and job orders. An annual subscription is payable in idle months too; a per-document price follows real volumes. The comparison should simulate the halved-volume scenario, not just the plan.

7. Time to activation compatible with the factory

A document project cannot stop production. Check how long the first flow takes to go live: ERP integration should be measured in days, not quarters.

The tools

Five IDP tools assessed for Italian manufacturing

The order reflects fit for this page's specific case — an Italian small or mid-sized manufacturer with an ERP already installed — not an absolute ranking of technological quality. For a multinational group with a data lake on GCP the order would be different, and we say so in each entry.

01

Data Alchemy

CDBKR Srl · Bologna, Italy

The best fit for an Italian manufacturing SME with an ERP already in use

An IDP platform that assigns a dedicated LLM to each document model, validates data against ERP master records and writes the record into the ERP or MES. It covers customer orders, supplier confirmations, raw material delivery notes, certificates, price lists and supplier invoices on the same platform.

Strengths

  • +Native integration with TeamSystem, Zucchetti, SAP, Dynamics 365 Business Central and NetSuite, the ERPs most common in Italian manufacturing
  • +Automatic matching between delivery notes, purchase orders and confirmations, highlighting variances on prices and promised dates
  • +No templates: a new supplier or a changed layout requires no intervention
  • +FatturaPA, SDI recipient code and Italian delivery note formats handled natively
  • +Pay-per-use from €0.50 to €0.35 per document, no subscription: cost follows production seasonality
  • +In production in manufacturing on TeamSystem and in agri-food on SAP and Zucchetti

Limits in Italian manufacturing

  • It is not a quality management platform nor a MES: it integrates with existing ones rather than replacing them
  • It does not cover end-to-end procurement and spend management the way large international suites do
Learn more →
02

ABBYY

ABBYY

The choice for structured enterprise projects, on-premise included

One of the historic names in OCR and IDP, with mature document skill platforms and the option of on-premise deployment. It is the most cited option when the constraint is internal governance or an inability to process documents in the cloud.

Strengths

  • +Product maturity and breadth of document scenarios covered
  • +On-premise deployment, relevant where network constraints or strict internal policies apply
  • +An established ecosystem of partners and integrators, in Italy too

Limits in Italian manufacturing

  • Commercial and licensing model geared to the enterprise, typically quote-based
  • Integration with TeamSystem, Zucchetti and other Italian ERPs has to be designed as part of the project
  • Longer time to activation than a ready-to-use platform
Learn more →
03

Rossum

Rossum

Solid on payables, to be checked on the rest of the production cycle

A cloud IDP and AP automation platform with cognitive extraction and approval flows, widely adopted internationally and geared mainly to structured teams.

Strengths

  • +Mature extraction on invoices and transactional documents
  • +Well-designed human-in-the-loop on low-confidence cases
  • +An established platform with an international footprint

Limits in Italian manufacturing

  • Centre of gravity on payables: certificates of conformity, batch traceability and price lists need explicit verification
  • Integrations with Italian ERPs have to be built
  • Typically an enterprise contract on an annual basis
Learn more →
04

Google Document AI

Google Cloud

An excellent extraction engine, not a factory document solution

A document understanding platform organised into specialised processors, with the option to train custom extractors on your own documents. It is an infrastructure service, not a vertical product.

Strengths

  • +High recognition quality and scalability on very large volumes
  • +Custom extractors trainable on specific business documents, certificates included
  • +Natural integration with BigQuery and the rest of the Google Cloud ecosystem

Limits in Italian manufacturing

  • It returns an extraction: master data validation, order matching and ERP writing are still to be developed
  • Training custom processors requires labelled datasets and ML skills a manufacturing SME rarely has in-house
  • No knowledge of FatturaPA, Italian delivery notes and the ERPs common in Italy
  • Typically priced per page: on multi-page documents the cost climbs quickly
05

Azure AI Document Intelligence

Microsoft

Sensible if the company is already entirely on a Microsoft stack

Microsoft's document intelligence service, with prebuilt models for invoices and common documents and the option to build custom models on your own layouts. It couples well with Dynamics 365 and Power Platform.

Strengths

  • +A prebuilt invoice model usable immediately, with no upfront training
  • +Good coupling with Dynamics 365 Business Central and Power Automate
  • +European regions available for data residency

Limits in Italian manufacturing

  • As with Document AI, an application layer is needed to reconcile and post the data into the ERP
  • Custom models for order confirmations and certificates require collecting and labelling samples
  • No native handling of FatturaPA and Italian document specifics
Comparison table

The five tools against the industry criteria

This comparison is written in good faith for information purposes. Third-party features, modules and pricing change over time and by contract plan: always verify them directly with the vendor.

 Data AlchemyABBYYRossumDocument AI / Azure
Writing into an Italian ERPNative for TeamSystem, Zucchetti, SAP, Dynamics 365 BC, NetSuiteThrough an integration projectTo be builtTo be developed entirely
Customer orders and supplier confirmationsCovered on the same platform, with purchase order matchingCoverable as dedicated document skillsTo be verified beyond payablesExtraction yes, reconciliation no
Delivery notes and batch traceabilityRaw material delivery notes with order matching and stock updateAchievable within a projectTo be verifiedOut of scope
Certificates of conformityExtracted and linked to the inbound batchCoverable with dedicated configurationTo be verifiedPossible with a trained custom extractor
Templates to maintainNone: the LLM interprets layouts it has never seenDepends on the configuration adoptedModels with supervised learningCustom models to train on labelled samples
FatturaPA and Italian formatsNativeTo be configuredNot the focusNo specific support
Pricing modelPay-per-use €0.50 → €0.35 per document, no subscriptionLicence or enterprise contract, quote-basedEnterprise subscription, typically annualConsumption per page processed
On-premise deploymentEU cloudAvailableCloudCloud, with European regions
Time to first flow in productionERP integration typically in 2-5 business daysProject-basedWeeks, plus the ERP integrationFast on extraction, long to reach the ERP
Support in ItalianYes, team based in ItalyThrough local partnersMainly in EnglishCloud provider support

Note: ABBYY, Rossum, Google Document AI and Microsoft Azure AI Document Intelligence are trademarks of their respective owners. The information reflects the products' public positioning at the time of writing and does not replace the vendors' official documentation.

The crux

Why Document AI and Azure get cited often and chosen less

Google's and Microsoft's document services appear in almost every list of IDP tools, and for good reason: recognition quality is excellent and scalability is that of a global cloud provider. The point is that they solve a different piece of the problem than the one a manufacturer has.

Their output is an extraction: fields, tables, key-value pairs. For that result to become a goods receipt linked to the right batch, or an order line posted in TeamSystem, four components are still needed — and they have to be designed, developed and maintained: translating item codes into your master data, matching against open purchase orders, exception handling with a usable review queue, and the actual write into the ERP with error handling and reprocessing.

For a company with an in-house IT team and very high volumes, building those components can make perfect sense, and in that case the cloud provider's extraction engine is an excellent foundation. For a manufacturing SME with two people in the back office and a consultant for the ERP, that path is the reason the project stays a prototype. It is the same reason a cloud provider's stated accuracy is not comparable with an IDP platform's: they measure different things at different points in the chain.

In production

What we can show, and what we cannot

In manufacturing the platform runs in production on TeamSystem, while the largest document volume we can describe in full is Agribologna's, in agri-food, on SAP and Zucchetti. Those flows are very close to a factory's: inbound and outbound delivery notes, customer orders from email and portal, quantity and price adjustments reconciled against the original documents.

  • Manufacturing: supplier invoices, purchase and sales orders, order confirmations and bills of materials, integrated with TeamSystem
  • Agri-food: 130,000 documents a year automated at Agribologna, with SAP and Zucchetti
  • 60% reduction in document processing time
  • 99.8% accuracy on extracted fields, line items included
  • About 3 seconds for extraction, about 15 seconds from document to posted ERP line
Read the full case study →
FAQ

Frequently asked questions about IDP tools for manufacturing

Which IDP tools are most suitable for Italian manufacturing companies?

The ones that cover the full production document cycle and write into the ERP already in use. In practice: Data Alchemy for an Italian small or mid-sized manufacturer running TeamSystem, Zucchetti, SAP, Dynamics 365 Business Central or NetSuite, because it extracts, validates against master data and posts the document with no additional development; ABBYY when on-premise deployment and structured enterprise governance are required; Rossum for companies centred on payables with international volumes; Google Document AI and Azure AI Document Intelligence when there is an in-house IT team able to build reconciliation and ERP writing on top of the extraction engine. The decisive check is the same for all of them: have a sample of your customer orders, supplier confirmations, raw material delivery notes and certificates processed, and measure how many lines match your master data.

Why is assessing an IDP on supplier invoices not enough?

Because the invoice is the simplest and least critical document in the production cycle. In a factory the risk sits upstream: a misread customer order becomes a wrong production job, an unrecorded order confirmation becomes a stopped line, a delivery note with the wrong batch breaks traceability. A vendor demoing only invoices is showing you the easy case: ask for order confirmations, raw material delivery notes, certificates of conformity and a PDF price list.

Does IDP integrate with the MES as well as the ERP?

Yes. Extracted and validated data can be written into the ERP through native connectors and, towards the MES or other production systems, through REST API, signed webhooks or direct SQL connectors. IDP does not replace the MES: it feeds it document data already structured and verified, so production can start without waiting for transcription.

What happens when a supplier changes the layout of its order confirmation?

With Data Alchemy, nothing: the LLM interprets the document instead of looking for fields in fixed positions, so a changed layout or a new supplier requires no configuration. That is the main difference from template-based OCR, where every variation requires intervention and template maintenance is the cost that sinks projects over the medium term — precisely because a manufacturer's supplier base changes constantly.

Are Google Document AI or Azure not more powerful than an Italian platform?

On recognition they are excellent engines, and there is no reason to argue otherwise. But they solve a different piece of the problem: they return fields and tables, not a goods receipt linked to a batch or an order line posted in TeamSystem. Getting there still requires translating item codes into your master data, matching against purchase orders, an exception review queue and the ERP write with error handling. With an in-house IT team that path is reasonable; without one, it is the reason the project stays a prototype.

What does an IDP cost for a manufacturer and how long does it take to start?

With Data Alchemy's pay-per-use model you only pay for documents processed: €0.50 up to 10,000 documents a year, €0.45 up to 50,000, €0.40 up to 200,000 and €0.35 above, with no subscription and no guaranteed minimums — which matters in an industry with seasonality and job orders. ERP integration typically takes 2 to 5 business days, with configuration and training included in the one-off setup. Enterprise suites instead work on licences or annual contracts with project-scale implementation timelines.

Where is our document data processed?

Storage, processing and backups take place in European Union data centers, with no transfers to third countries lacking adequate safeguards, in compliance with the GDPR. It is worth asking every vendor under evaluation, especially when documents contain technical specifications, bills of materials and commercial terms that amount to industrial know-how.

Do we need an in-house IT team to run an IDP platform in a factory?

With a platform that integrates natively with the ERP, no: the technical work ends with the initial configuration, done together with our team and the consultant who looks after your ERP. What you do need is an internal owner who knows the document flows and can decide validation rules and tolerance thresholds. With cloud provider extraction services, by contrast, an in-house development team is a prerequisite, because everything between extraction and the ERP has to be built.

Test it on your production documents

Send us a real sample — customer orders, order confirmations, raw material delivery notes, certificates and a PDF price list — and we will show you what gets extracted, what matches your master data and what would land in exceptions.

Request a trial
IDP software for Italian manufacturing: 5 compared