Turn your idea into a real product with AI. In Lugano.

Swiss AI builders, based in Lugano. We build startup products, AI agents and on-premise local AI for startups, SMEs and regulated teams.
Senior rates, a fraction of the hours.

Book a 30-min build call → See what we've built
01 What we do

AI development:
idea to shipped product.

No quarter-long discovery phases, no 80-page decks. We compress scoping, build and deploy into a four-week cycle that ends with working software in your hands, and the code that runs it after we leave.

01 build plan

Scope

Up to a week with you. We cut the idea down to the version worth building first: what ships now, what waits, what it takes. You leave with a concrete build plan and a number.

OUTPUT · BUILD PLANWEEK 01
02 the real thing

Build

By end of week 2 you're clicking a real prototype on real data. Then app, backend and infrastructure, built by senior engineers: production-grade code, hosted in Switzerland or your cloud of choice.

OUTPUT · DEPLOYEDWEEKS 02–04
03 yours to run

Hand over

Code, docs and a walkthrough, so you or your team can extend it without us. No black boxes, no vendor lock-in, no surprise bills next quarter.

OUTPUT · AUTONOMYWEEK 05+
The full build process→
[ Policy 001 ]

If it doesn't work, you don't pay.

How that works→
02 Client work

Built for clients.

Production systems we shipped for other companies - the same loop, applied: scope, build, ship. From first call to handover.

01 smemarket.ch Swiss SME Marketplace → 02 Know-Your-Client Person Due Diligence → 03 M3 Risk Hedge Fund Performance & Risk →
M3 Risk preview

Operator-grade dashboard for crypto options trading desks. Black-Scholes pricing, real-time Greeks aggregation (Delta · Gamma · Theta · Vega · Rho), margin requirement + liquidation monitoring, stress testing, and concentration risk analysis. Built for hedge funds running discretionary or systematic BTC/ETH options books - one source of truth across exposure, risk, and performance.

The desk's live risk view: Greeks, margin and liquidation limits watched in real time, every trading day.

STACK DjangoSQLiteCoinGeckoBlack-ScholesVanilla JS
04 xo8.media AI-powered media agency → 05 yeswetrust App to match founders with funders →
03 Incubated startups

Our own bets.

Products we incubate ourselves - same playbook, except it's our own money on the line.

04 The team

Boutique. By design.

We don't have account managers, junior associates, or a sales pipeline. The people you meet on the first call are the same ones writing your code.

M3 / 01
Paolo Montemurro

Paolo Montemurro

Founder & CEO

STACK
vision · ai · strategy
LOG
building the future, one agent at a time
M3 / 02
Simone Caggese

Simone Caggese

Partner & Cyber Security

STACK
cyber security · infrastructure · fullstack
LOG
if there's a way in, I'll find it first
M3 / 03
Riccardo Giordano

Riccardo Giordano

Implementation Wizard

STACK
backend · infra · apis
LOG
turning ideas into production
M3 / 04
Edoardo Schiatti

Edoardo Schiatti

AI Solution Engineer

STACK
ai · integrations · engineering
LOG
from problem to production
M3 / 05
Oliver Camponovo

Oliver Camponovo

Partner

STACK
strategy · partnerships · growth
LOG
opening the doors that matter
05 Questions

Asked before
every build.

How much does an AI project cost in Switzerland?

We publish what is fixed: a custom web app built with AI starts at CHF 10'000, a scope we used to quote at three times that, and on-premise local AI installs start at CHF 3'500. Every build is scoped in week one, and you leave scoping with a concrete build plan and a number, before committing to anything.

How does “if it doesn't work, you don't pay” work?

During scoping we agree in writing what “working” means for your build: the concrete things the software must do at delivery. If what we hand over doesn't do them, you don't pay for the build. The definition is fixed before we start, so there is nothing to debate after.

Can AI run fully on-premise, without our data leaving the building?

Yes. We install open-weight models on hardware you own: private, offline-ready, with a chat interface for your team. Nothing is sent to a cloud provider, which is what makes it workable for fiduciaries, finance and healthcare.

How long does an AI build take?

Four weeks is the standard cycle: scoping in week one, a real prototype on real data by the end of week two, production deploy in weeks three to four, then handover, code, docs and a walkthrough, until your team can extend it without us.

Who do you work with?

Three kinds of clients: startups that need a first product built, SMEs that want AI agents and automation in their workflows, and regulated teams, fiduciaries, finance, healthcare, where data must stay on-premise.

Do we own the code?

Yes. Code, documentation and accounts are yours at handover. No black boxes, no vendor lock-in, and pre-existing Mont3 tooling never holds your product hostage.

06 Get started

Let's ship the first version this month.

[ 01 / Recommended ]

30-minute build call

Free. You have the idea: we cut it down to the first version worth building, tell you what it takes and what it costs. No sales script.

Book the call→
[ 02 / Not sure yet ]

Don't know what to build?

Free audit: we look at how you work, find where AI actually pays for itself, and tell you honestly whether it's worth building at all.

Book the audit→
[ 03 / Or just email ]

Tell us the idea

Send a few sentences about what you want to build, or the process eating your team's time. We'll reply within one business day with first thoughts.