WORK, PROTOTYPES AND DEMOS
Some are finished work, others show how we approach a problem. Every card says clearly which kind it is.
The demos don’t collect requests or take payments. The demo websites belong to fictional Italian businesses; the control room demo is available in Italian only.People looking for a place to stay decide from the pictures, then have to call to find out about rooms, dates and availability.
The story of the place, rooms and spaces you can browse, and a stay request that already arrives with dates and number of guests.
A high-end property is presented in the same language as every other one, and the difference doesn’t show.
The same property in a second visual direction: suites, architecture and details, with a more restrained presence.
The winery can be visited, but online there are only labels: people who would like to come can’t find how to ask.
Land, wines and people in a single path, four palettes, two languages and the tasting request at the end of the story.
Bottles are only sold at the winery, and only during opening hours.
Catalog, product pages and cart with a boutique presentation. Payment is simulated: it is connected and tested during the project.
Guests travel from far away and the practical information gets lost between messages and phone calls.
A complete event website: opening video, story, program and RSVP with plus-ones. No response is sent or saved.
Replies arrive one at a time, and every guest comes with plus-ones and dietary needs.
A simple invitation, the program of the day and a reply form for each guest, with plus-ones and diets.
No two weddings are alike: the same structure, with another identity, has to be able to change completely.
A bright, botanical take on the same path: story, program, practical information and RSVP.
Someone looking for a contract manufacturer wants to know what you can machine and to what tolerances, not what you sell.
Shop floor, machines, tolerances and dimensional inspection, with a quote request that asks for the process and the tolerance.
The customer knows the part number and wants to compare two components, but has to write and wait.
Search by part number, side-by-side technical comparison and a downloadable specification to attach to the quote request.
A mountain product is explained by its aging and its pastures, but online it is just a list of names.
Cheeses and cured meats you can filter, the story of the aging, the shop and cellar visits with an availability request.
Direct sales and on-site experiences live on two different channels, and neither of them is the website.
A shop with a cart and a calendar to book tastings, within the same path.
Every quote starts with a site visit, but first you need the opening type, material and measurements: that takes three phone calls.
A configurator for opening, materials and measurements that prepares the site-visit request already filled in.
Project status, decisions taken and companies to contact live in different files, and to know where things stand you have to ask someone.
A dashboard that reads the working files: projects with their open questions, decisions with their constraints, proven capabilities, leads with their record and the contact log. Numbers nobody has recorded stay empty.
A spreadsheet collected by hand arrives with duplicates and invalid amounts, and nobody notices before the total.
Raw sheet → cleanup → checks → summary, run in the browser. Next to it, a comparison between two databases and the routing of a request.
The design is finished in the model, but drawings, schedules and Excel sheets are copied by hand: every change restarts the transcription, and a wrongly copied value is found late.
Four automations inside the design software already in use, built by one of the founders: they read the model and write drawings, schedules, installation sheets and Excel on their own, regenerated at every change.
Every job has new measurements, but the rules to draw it are always the same: they live in a spreadsheet or in one person’s head, and are recalculated by hand every time.
Dedicated software, built by one of the founders: it takes the parts list, rebuilds the modules, calculates the fixings with safety margins and generates dimensioned CAD drawings, a cut list and labels.
Turning an order into crates and fitting them into the container is done by hand, with rules one person knows: the oversized package or the half-empty container shows up at loading time.
An engine that applies the packing rules, calculates the packages, fills the container floor, flags oversized items and unknown codes instead of losing them, and proposes the right container, with a top view for whoever decides.
When AI enters a complex business process, handing everything to a single assistant makes it hard to know who did what, with which information, and where to step in before a mistake.
We built an architecture in which specialized agents carry out distinct tasks, pass each other verifiable outputs and ask for human approval at the critical steps.
Procedures, decisions and constraints can be spread across different documents or even contradict each other. An agent needs to know which source prevails and which actions it cannot take.
We built a governed memory for agents: structured documents, a hierarchy of sources, derived indexes and automatic checks make sure information and constraints are applied consistently.
When market information comes from different sources, collecting a lot of it is not enough: you need to know what it actually supports, how complete the research is and where the evidence is weak.
We built a pipeline that gathers different sources, measures how well the research covers the question, keeps the evaluation criteria separate and brings the result to a human decision instead of hiding the uncertainty.
Producing analyses and documents with AI is easy; checking that every claim is actually backed by the sources takes a research and review process you can control.
We built a pipeline that researches and weighs the sources before writing, links claims to evidence and uses separate agents to critique, verify and re-examine the result before it is closed.
Turning technical documents, research or verified material into shorter content can introduce simplifications, claims stronger than the original or information out of context.
We built a system that keeps every piece of content linked to its source, checks how claims are transformed and flags or blocks those that go beyond the defined criteria.
When many pages, product sheets or documents must respect precise sources and rules, manual review gets slow and can let through exactly the claims that matter most.
We built a system that extracts claims in a structured way, links them to their context, compares them with reference sources and measures its performance against independent annotations.
Documents, management software, Excel and email each hold a piece: nobody has the picture of what exists, what is true right now and what is changing, and the status is rebuilt by hand.
We built a system that keeps an up-to-date model of the company and also knows how it is running itself: which sources it reads, what it is allowed to do, what it costs and when to stop and ask a person. Complete and tested, not yet installed at a client of the studio.
The procedure exists, but only one person knows it: everyone else interrupts her or searches through folders.
Three questions, three answers with a source to open and check. On the third, the assistant says it doesn’t know instead of making something up.
An out-of-range reading is discovered from the damage, and a sensor that stops responding looks like a room with no problems.
Four settings (home, greenhouse, farming, industry) with temperature, humidity, controls, thresholds and loss of connection.
Cameras record, but the footage is only watched after a problem; and sending it to a cloud service means spaces and people leave the building.
A vision system that runs on a computer inside the company: it recognizes people, gives each one a number that holds even behind an obstacle and records the session, without sending video outside.
How a component is made becomes clear only when you hold it: a photo doesn’t show where the seals are.
An explorable 3D model with an exploded view and component details, next to visuals of the same product.