EP.005 – My Coffee for the Maikers Event

This article contains the full transcript of the speech delivered at the “AI is great, but how do I use it in my business” event organised by the Maikers community.
A presentation stemming from a real-life case of innovation in a company: the AI agent EMMA, to reflect on how to connect processes, evaluate cost-opportunity, and build sustainable and measurable optimisation paths.

---

Full Transcript

Moderator: Let's move on to the second case study. Massimo Gobbo, strategic marketing consultant, will tell us about a real-life case. Let's start looking at practical case studies.

---

Massimo: Thank you, thank you. Good evening everyone, good afternoon. I'll start by thanking those who came before me, for two reasons fundamentally.

The first is that when you run a relay race, and ideally we are running this relay, the first person to start, the first 100 metres, never makes it into the Gazzetta dello Sport the next day, because there's never a photo of the one who starts. However, if the one who starts badly or stumbles or drops the baton, I think even those who come after will struggle to win a medal. I believe the 100 metres with which we started this race are 100 metres with a very good intermediate time and which also leave us with many insights.

That's why I start precisely from one of the last elements Alessandro shared: before building the great AI pyramids within the company, select a project, even a small one, select the process to optimise, select the tool or tools to use, and start from there, iterate, start improving the process.

Why? Because what actually happens in many, even medium-to-large companies, and also in small ones?
Female and male entrepreneurs go to an artificial intelligence event on Friday afternoon.
On Saturday morning, the entrepreneur is in the office, thinking about it.
On Sunday afternoon, they're on their phone looking for information.
On Monday morning, they wait to call the person who manages their server, the one who redid the CRM, the friend “expert in these things”. But after all these phone calls and meetings, what happens is that the project that emerges is this:

“Let C be a disk of radius R inscribed in the base of a rectangular parallelepiped with dimensions L × L × H, where L is equal to 2R”. And what do we want to do with all this stuff? Well, we want to “create an automatism to verify the continuity between the radial arrangement of the sectors and the structural orthogonality of the box containing our disk.”

When an entrepreneur, or even a single professional, arrives at this description, they say:
“Alright, let the mega-company with 5,000 employees do the artificial intelligence. If we try to do this, we'll lose our shirts.”

The point is that all this description actually spoke of a pizza inside its box, a disc divided into triangles inside a parallelepiped. Very often what happens is that instead of a task force that selects the project, what a former colleague of mine called the Office of Complicating Simple Affairs (OCSA) is created.

And that's precisely what we must avoid when trying to innovate or bring something new into the company.

On this, I asked the generative AI of the Alibaba group to create a Chinese proverb for me to reinforce this concept, and you can see it written:

“He who wishes to cultivate the forest must first plant a single seed and let the wind not cover it with useless leaves.”

As I say there, I didn't write this, Confucius didn't write it, no great philosopher wrote it… a generative AI wrote it. But I'd say it hit the nail on the head. I'll keep it in mind.

---

So, my case study I'm going to talk about: as mentioned in the presentation, I deal with marketing strategy, product launches, and growth in international markets.

I was called by a B2B manufacturing company from Veneto that produces mechatronic components and sub-supplies. It produces components both in-house, with an internal production department in Veneto and abroad, and commercialises products from other companies.

When they called me, they asked for three things:

1. improve the commercial performance of the internal team;
2. create the conditions to enter European markets where they were not present;
3. increase the number of incoming leads.

We started discussing with all departments: sales, administration, production, and logistics.

As a marketing consultant, one of the first things I do is create a new lead generation campaign, select new area managers in the new markets, and optimise the entire process.

This worked: new contacts arrived, samples were sent, the first small orders were generated, revenue started to come in, but soon, in an update meeting, it emerged that all this activity was causing great confusion at the documentation, warehouse, and monitoring levels.

The administration said:

“In just one month, I have three unclosed consignment documents, two sample documents for destructive tests returned… this is going nowhere.”

Logistics added:

“I can't get to 6:30 in the evening with the sales rep calling me: 'Tomorrow morning that part must be in Finland, otherwise we'll be sued,' and I don't even know where it is.”

---

What did we imagine at that point?
We imagined hiring Emma at the company.
Emma became the AI colleague within the team.

Let me open a parenthesis: why Emma, and why does she have that hair?
It all started with an anecdote at the coffee machine with the entrepreneur: I showed him a photo of my dog that I had ChatGPT process. He said: “I want to try it myself with my dog, named Emma.” He has a Maltese named Emma and a blonde girl with glasses came out. From there, “Emma” was born, who then became the virtual colleague of the company.

---

What does Emma do? She integrates with company processes and supports the different departments to avoid the problems I described earlier. The administration said a key phrase:

“Emma has humanised the management system.”

The company had changed its management system a year and a half earlier, so hearing “Emma has humanised the management system” almost moved us.

---

Emma works like this.
External agents and internal sales reps activate her via a Telegram bot.
They fill in fields (product code, specifications, etc.).
Emma verifies if the code is in the master data, replies “ok, received, we can proceed” and creates a unique ID for the sample or prototype, saving it in a shared database.

Meanwhile, she updates logistics, production, administration, and sales that the process has begun.
Then, through the management system's APIs, she communicates with the ERP, monitors the progress of the production or retrieval process, and pre-fills a transport document, which is validated by administration.

The second phase is the one that saves the most time: whereas before sales reps used to call to ask “what's the status?”, now a simple message on Telegram is enough. Emma replies: “It's being shipped, it's being packed,” etc.

She is also able to retrieve updated customer data and inform departments. When production enters “product ready for shipment,” Emma (who knows dimensions and weight from the management system) updates administration and sends the sample ID. The sales rep can then notify the customer (“it will arrive in two days”) and logistics prepares the shipment.

---

The company is now working to ensure that:

• delivery notes no longer require manual approval;
• Emma automatically books the courier;
• Emma also becomes an interface with the customer (tracking, documents, shipping status).

This generates automatic responses, fewer conflicts, and fewer documentation errors, resulting in enormous time savings.

---

The owner wanted to calculate everything: how much time is saved, how many calls avoided, how many interruptions reduced. He defined “Emma's Golden Week” as the week in which the company saved a total of 12 hours of work. Not all weeks are like that, but the figure is still remarkable.

Important note: it's not free. Most of the queries and integrations are not free commands and the internal IT department had to work a lot on it. The cost-opportunity of the project holds up, but it shouldn't be approached with the idea that “AI is free anyway,” because it isn't.

---

Questions from the audience

Question from the audience: Was Emma developed by departments already present in the company or did you have to hire someone?

Massimo: Yes, it was developed internally.

Question from the audience: Can you say what tech you used for this project?

Massimo: For the very first version of Emma, they used n8n to control everything else.
Then, given the validity of the flow, they coded the internal part. Integration with the ERP, use of LLMs for reasoning, initially relying on open ChatGPT, and now they are testing solutions for managing foreign languages, because Emma will also have to respond to foreign customers.

Question from the audience: Using an LLM, how did you manage hallucinations and response errors?

Massimo: This is precisely why the documentation, even if pre-filled, is always checked manually. At first, there were many errors, then they were reduced to almost zero. It's a bit like when we first used Excel: it doesn't make mistakes, but if you make a mistake, it returns a date from 1970! Today, we're in the order of very few errors.

Question from the audience: Was Emma born as a planned project or as a need?

Massimo: In reality, it was born as a need, not as a planned project. The company had already tried in the past to assign that role to internal people, first to administration, then to sales, but it hadn't worked. When the company grew internationally, the first idea was “let's get an intern for the back office.” The risk was creating another bottleneck. That's how Emma was born: to automate without adding bureaucracy and to keep track of information objectively.

---

If you enjoyed the presentation and want to continue discovering other stories and strategies from the world of European SMEs, subscribe to my LinkedIn newsletter Marketing Café B2B. You'll find new interviews, practical insights, and ideas to put into action immediately.

.libutton { display: flex; flex-direction: column; justify-content: center; padding: 7px; text-align: center; outline: none; text-decoration: none !important; color: #ffffff !important; width: 200px; height: 32px; border-radius: 16px; background-color: #0A66C2; font-family: "SF Pro Text", Helvetica, sans-serif; }

Subscribe on LinkedIn