Intelligent automation: where to start in your company

To start automating with AI, choose processes with high volume, high repetitiveness, and low variability. Avoid processes that require too much human judgment. The best candidates are order management, email triage, reporting, and data entry — tasks where AI generates measurable ROI within 30 days.

The infinite pilot project trap

Many Italian SMEs find themselves stuck in a phase of "perpetual exploration." They understand that AI and automation can make a difference, but they struggle to move from interest to concrete action. The result is a series of pilots that never scale, demos that never become production, investments that never generate measurable ROI.

The problem, almost always, is not the technology. It is choosing the wrong processes to automate.

The framework for identifying the right processes

Not all processes are equal when it comes to automation. Some lend themselves naturally, others require too much flexibility or human judgment to be profitably automated today. Here is how to distinguish between the two.

The four evaluation criteria

• Volume: Does the process repeat many times? A process that occurs 500 times per month has very different automation potential than one that occurs 10 times.
• Repetitiveness: Are the rules governing the process clear and relatively stable? Or do they continuously require contextual judgment and exceptions?
• Current process cost: How many person-hours per month does this process absorb? What is the cost of an error?
• Data accessibility: Does the process produce and consume structured data, or does it primarily operate on unstructured information (emails, PDFs, conversations)?

The 5 processes to automate first in an SME

Based on dozens of implementations in Italian SMEs, these are the processes that offer the best impact-to-complexity ratio:

1. Email and inbound communication management

The company inbox is often the most underestimated bottleneck. Automatic classification, routing to the right people, and draft response generation — these three elements alone can recover 1-2 hours per day per person.

2. Order and commercial document management

Receiving orders via email or portal, checking availability, order confirmation, ERP update. In many manufacturing SMEs, this process occupies 2-3 FTEs. With automation, it can be managed with minimal oversight.

3. Reconciliation and data entry

Supplier invoices, expense reports, warehouse movements — manual data entry is expensive, slow, and error-prone. AI document processing has reached accuracy levels that exceed human input.

4. First-level customer service

70-80% of customer service requests fall into 10-15 recurring categories. An AI agent trained on your historical data can handle these requests autonomously, escalating to a human only for exceptions.

5. Reporting and KPI monitoring

Collecting data from different systems, building periodic reports, sending alerts on anomalies — all of this can be automated, freeing management for decisions instead of data gathering.

How to build the business case

Before any investment, it is essential to build a quantitative business case. The structure is simple:

• Current process cost: Hours/month x average hourly cost + error cost + opportunity cost
• Post-automation target: Residual hours/month x hourly cost + AI solution cost
• Annual savings: (Current cost - Post-automation target) x 12
• Payback: Implementation cost / Annual savings

In SMEs where we have implemented automation solutions, average payback is under 6 months for Tier 1 processes (high volume, clear rules).

Mistakes to avoid

Three frequent mistakes that compromise automation projects in SMEs:

• Automating broken processes: Automation amplifies what is already there. If the process has intrinsic inefficiencies, automating it amplifies them. First optimize, then automate.
• Starting too big: The most effective automation project is often the simplest. A workflow that works and generates real value is infinitely more useful than a complex system that never goes to production.
• Underestimating training: The staff who will use the automated system must be involved from design, not just from implementation.

The next step

If you have read this far, you probably already have 2-3 processes in your company in mind that could benefit from automation. The next step is to sit down with someone who has already implemented similar solutions and do an honest assessment of the potential.

At IL DOGE DI VENEZIA we do exactly this: a structured assessment of your processes, a quantitative business case, and a realistic implementation plan. Tell us about your situation — the first conversation is free.