[ SOLUTIONS ] / [ AI ERP INTEGRATION ]

We put AI inside the ERP you already have.

Your ERP works, but manual interactions with it add up. We add an AI layer on top of the existing ERP without touching the core: TeamSystem, Zucchetti, IFAC, Mexal, custom systems. The AI runs in separate microservices that talk via API. The risk of breakage is structurally low.

Let's talk
[ THE PROBLEM ]

What happens today.

The company wants to use AI but can't replace an ERP that's been in production for years, holds all the historical data, is integrated with accounting and logistics, and works. The binary 'all AI, all rebuilt' decision doesn't stand up.

At the same time, the functions where AI makes the most sense (document extraction, email classification, operator support, report generation) are exactly the ones in contact with the ERP. Without access to its data, AI lives in a separate silo and isn't useful.

The realistic path is middle-out: an AI layer that talks to the ERP via API, without modifying the core. AI functionalities run in independent microservices. The ERP is left untouched. If tomorrow the AI layer changes or is decommissioned, the ERP won't even notice.

AI doesn't replace the ERP. It sits on top, and talks.

[ HOW IT WORKS ]

The solution, broken into parts.

  • Separate AI microservices

    AI functionalities run in microservices independent from the ERP. A bug blast radius is limited to the microservice; the ERP stays operational. AI updates without freezing the business.

  • Integration via standard APIs

    Read/write via REST, GraphQL, SOAP APIs or dedicated connectors for common Italian ERPs (TeamSystem TS Digital, Zucchetti HR, Passepartout, Mexal). For custom ERPs we build bespoke middleware.

  • Modular use cases

    Start with one or two high-ROI cases (e.g. invoice extraction, email classification) and expand progressively. Each case is its own microservice, switchable on or off independently.

  • Governance and audit log

    Every AI decision is logged in a structured way: rationale, input, output, confidence level. Ready for internal audit or DPO requests. AI Act compliance on track for August 2026.

[ WHO IT'S FOR ]

The typical profiles who benefit.

  • Italian SMEs on TeamSystem or Zucchetti

    You have active licenses, know the ERP, and don't intend to change it. You want to use AI where it actually helps: documents, classification, operator support.

  • Companies with a legacy custom ERP

    The ERP was built years ago, still in production, holds the tribal knowledge of the company. Replacing it would cost hundreds of thousands of euros and months of operational freeze. AI as a layer is the pragmatic alternative.

  • Professional firms with vertical software

    Software for accountants, law firms, labour consultants: AI helps on repetitive activities (PDF reading, case classification) without having to change the main working tool.

[ WHAT WE NEED ]

Transparency on what the client does.

Before we start we need a few accesses and decisions. All reasonable, no surprise asks.

  • Technical access

    • API access to the ERP (read for read-only cases, write for the ones that require it)
    • Development account with test data representative of the production set
    • List of ERP modules we will need to touch (master data, orders, invoices, accounting, warehouse)

  • Operational decisions

    • Priority list of use cases to implement
    • Operational boundaries (what AI can do autonomously, what always needs human approval)
    • Applicable compliance constraints (GDPR, AI Act, regulated sectors)
[ TIME AND COST ]

Indicative numbers, not quotes.

TIME
First use case integrated in 6-10 weeks. Subsequent expansions, 3-6 weeks per new case.
COST
Range €20,000-60,000 for the first case, €10,000-30,000 for each additional case. Operating LLM cost €100-1,500/month depending on volume.
MODEL
Time & material for the initial exploratory case, fixed milestones for subsequent expansions.

Indicative numbers. For an accurate quote, let's talk.

[ FREQUENTLY ASKED ]

Answers to the most common questions.

Do I have to replace my ERP to integrate AI?

No. Our approach is the exact opposite: AI lives in separate microservices that talk to the ERP via API. The ERP is left untouched, AI features are switchable on and off independently, and your existing investment is preserved.

Which Italian ERPs do you have direct experience with?

We work with TeamSystem (TS Digital, Polyedro), Zucchetti (HR, Ad Hoc, Infinity), Passepartout (Mexal, Welcome), and many custom ERPs built over the last 20 years. For uncommon or custom systems we build dedicated middleware.

How much does keeping the AI layer in production cost per month?

Typically €200-2,000/month, made of: LLM cost (€100-1,500 depending on volume), microservices hosting (€50-300), monitoring (€50-200). Cost varies a lot with the AI-operations volume. On-prem models for high volumes cut the variable cost.

Is the data sent to the AI safe?

We work with enterprise providers (OpenAI, Anthropic, Mistral) that offer no-training-on-customer-data options, or with on-prem / private-cloud models for very sensitive data. All data is encrypted in transit and at rest, access log is structured, audit trail continuous.

[ LET'S TALK ]

Recognize your case?

Write a couple of lines about your context. We'll reply within 24-48 hours with an initial assessment and a first orientation on time and cost.

Let's talk
[ RELATED SOLUTIONS ]
[ WHERE IT WORKS BEST ]
[ PARENT PILLAR ]
AI agents and LLM integration We don't replace your systems. We give you AI agents that integrate with them.