From “only Raffaele knows” to “the Company Brain knows”
How M3SN, the exclusive Tohatsu importer for Italy, put its technical and commercial knowledge in the hands of its entire network. And started producing technical training with AI, too.
Knowledge is the product as much as the engine.
M3SN doesn’t just sell engines. It has to keep an entire commercial network running (agents and more than 140 dealers spread across the country), supplying price lists, spec sheets, part numbers, ordering procedures, marketing materials, pre- and post-sales support, and training.
Every day, from all over Italy, dozens of precise questions come in:
And there’s a second front, stated by the company itself: training the network. Tohatsu’s service philosophy is before service, preventing problems, and it translates into regular courses and seminars so every dealer is trained directly. Producing that training, though, used to cost time and many different hands.
Knowledge lived in the heads of a few people.
For years, the answers always came from the same people. Technical know-how was held by a few internal references, ordering procedures went through one single person, and the four dealer price lists and seasonal promotions were easy ground to slip on.
- 01 The network waited. An agent or a dealer trying to close a sale on a Saturday afternoon had to wait until Monday for a part number or a confirmation of availability.
- 02 Key people were interrupted constantly. Internal experts spent their days answering the same questions, taking time away from higher-value work.
- 03 The risk of error was real. With four price lists, a public price and overlapping promotions, it took very little to quote the wrong list, or to share a net dealer price with the public.
- 04 Producing training was slow and costly. Building a technical course, with its slides, diagrams and quizzes, meant starting from scratch every time.
People become middleware.
It’s the problem we at ekory call context debt: when knowledge isn’t structured and shared, people become mandatory hubs every piece of information has to pass through. It’s not a technical nuisance. It’s a cost you pay every day: in time, errors and lost sales.
One Company Brain, many Ekos.
With ekory, M3SN built a Company Brain: a single knowledge base fed by the company’s real documents (price lists, Product Information Guides, manuals and homologation files, accessory catalogues, SWAN procedures, marketing materials) and made queryable in natural language.
On that base we configured specialized Ekos, each tuned to a precise audience and task. Many different assistants, a single source of truth.
Assistente Tohatsu
Builds complete, configured quotes, checks stock availability, finds part numbers even when superseded, assembles service kits, explains the warranty and the 70SE, guides ordering on SWAN, and compares models against the competition.
Avatar di Raffaele
The digital twin of one of the company’s most experienced people: repowering advice for the right hull, installation diagrams, NMEA 2000 networks, hydraulic steering. The whole network can ask Raffaele, without tying up Raffaele.
Service Ai
The technical assistant for whoever has their hands on the engine: compression ratios, valve diameters, oil pressures, torque specs, materials, sensors, thermostats. With a focus on generation-to-generation comparisons, useful for diagnostics.
Marketing Ai
Writes captions for reels, replies strategically to social comments, even the most critical ones, extracts technical data in structured form, and keeps the brand’s tone of voice consistent.
Esperto Liqui Moly
Dedicated to the Liqui Moly Marine line, also imported by M3SN, with order tables, discount tiers, rebates and specific technical support.
The technical-course factory.
There’s one use that, on its own, changes what the project means: with ekory, M3SN doesn’t just consult knowledge: it turns it into training. Exactly the network training that has always been part of its service philosophy.
Paolo Zumbo, the company’s technical reference, used ekory to build an entire course, “GT 30 & Tohatsu”, for the technical commissioners and race officials of competitive powerboating, using AI at every stage.
Structure from real sources
From the homologation files loaded into the Company Brain, the AI generates the course outline. Paolo reorganizes it in chat and the system regenerates it, keeping sources, versions and the ability to step back.
Custom technical images
Mechanism diagrams, the four phases of the 4-stroke cycle, P-V diagrams, a GT30 boat with a Tohatsu engine. All in one consistent style, archived in the course folders.
A reusable visual language
From a single reference image, the AI extracts a real template: palette, fonts, grid, iconography, style rules. Every following slide keeps the same language.
Slides and quizzes, ready
Lessons become ready-to-use slides, with the option to add quizzes for the final assessment. Work that used to take weeks now lives in a single flow, governed by the domain expert.
You ask, you get the answer, you get back to work.
The same questions that used to require a phone call now get answered in seconds. With the source cited, always.
· 40 Mega engine
· EPD required
· Transport bracket
· Recommended instruments and accessories
Reminder: the order must be entered on SWAN. Ready to send to the customer.
The most interesting part is what the system refuses to do.
Knowledge is shared, but with the right permissions, the right limits and full source traceability. That’s what we mean by a governance-first architecture.
The system also knows the unwritten rules: that the net dealer price isn’t shared with the public, that a model on its way out of the range shouldn’t be pushed for the long term. And when it gives a technical figure, it cites the source, so the reader can trust it and verify it.
Triple-digit growth across every indicator.
Over the last 30 days on M3SN’s Company Brain. The more the network finds useful answers, the more it comes back to ask: adoption that feeds itself.
| Eko | Conversations | Users |
|---|---|---|
| Assistente Tohatsu agents | 489 | 11 |
| Avatar di Raffaele dealers | 1,552 | 138 |
| Marketing Ai marketing | 149 | 4 |
Raffaele’s avatar is the assistant with the widest reach (138 people across the dealer network query it), while Assistente Tohatsu serves a smaller group of agents with a higher volume per user. Exactly the two audiences they were designed for. Activity concentrates during working hours, peaking on Wednesday morning, but evenings and weekends aren’t missing either: the tool has entered people’s real workflow.
Four foundational choices.
What made this project different from “a chatbot on PDFs.”
One base, several specialized assistants
Not a generic do-everything, but one Eko per audience (agents, dealers, workshop, marketing) and per task.
People’s knowledge, scaled
The avatar of a key person multiplies their expertise without multiplying their interruptions.
From consultation to creation
The same Company Brain that answers questions also produces the content: courses, slides, technical images, quizzes.
Governance first
Permissions, limits on what’s sensitive, and source traceability: sharing knowledge without losing control of it.
“Training is sacred, but perfect memory is a mirage. No dealer remembers every part number, price and procedure for one brand among many, while they work. Ekory doesn’t replace the courses, it brings them to life every day: the right knowledge, at the right moment. It’s not about selling more, it’s about always being there for the network. And that’s how a brand becomes a leader.”
Knowledge has stopped living in the heads of a few people.
Now it lives in the Company Brain, and it belongs to the whole network. Want to see what it could do for your company? We take your materials, build an Eko in a day, and show you, side by side, what changes.