AI Skin
AvailableYour rules, tasks and permissions, installed in the AI assistant your teams already pay for. One skin covers development, finance, sales or support.
For companies where five people do the same job five ways.
Learn moreWhere the routine ends, you begin
We do one thing
AI that works inside your company.
- 26real projectsin build or production
- 3productsAI Skin, agents, Atalaia
- 2delivery partnersengineers on the ground
CH.01 · Three ways to start
Three products open the door. One team does the work.
Each product has its own site. Behind all three sits the same engineering team, and most engagements start with one of them.
Headless Agent Builder
Building nowRoutine work that runs without a screen. An event, a schedule or an API call starts it. The cases that need a person reach one; the rest post on their own.
For the work nobody should do by hand at three in the morning.
Learn moreAtalaia
Transversal · cuts across bothSee what your AI agents are doing. Every automation and AI worker you run, in one record. Declared on day one; connected when you choose.
CH.02 · Real projects
We are building twenty-six projects. Six of them are here.
We name each client by sector rather than by name, and we describe the outcome in plain words. Every number here was measured.
CH.03 · How we work
Discovery, proof of value, production. Then again.
Every customer runs the same cycle and sees results in weeks. Each round starts from the numbers the last one produced.
Then round again, with what we measured.
- 01
Discovery
two weeksWe write the problem in one sentence and look at your data. You get one prioritised backlog and a clear go or no-go.
- 02
Proof of value
weeksA working agent or skin on your data. Failure cases written down before anyone trusts it.
- 03
Production
installedOn your systems, under your permissions, with a trail. We stay until it runs without us.
- 04
Measure
alwaysWhat changed, in words and in the numbers you already track. The next round starts here.
CH.04 · Where it runs
Where it runs is a decision we make with you.
The same build runs in three places. Your compliance rules, your data and the tools your teams already have decide which.
Your assistant
Inside the one your teams already use
It works inside Claude, Codex or Copilot, so nobody has to open a new application. If you do not use an assistant yet, we choose one with you.
Local models
On your hardware, when compliance requires it
Open models, installed and tuned inside your perimeter. Nothing leaves the building.
Cloud, on our terms
Through OpenRouter, only models with no data retention
Pinned to the region your data must stay in. Europe by default.
Where your data may go is settled with you before the first line of code.
CH.05 · Who we work with
Partners who deliver with us. Clients who run what we built.
Two partners put our engineers on their projects. Two clients run what we built. The rest are teams we have worked with.
Puts our engineers on its projects, on the ground with the client.
Puts our engineers on its projects, on the ground with the client.
Runs one AI Skin across marketing, sales and finance, in production.
Builds AI into its accounting platform, engineered with their team.
- 2partners who deliver alongside us
- 26projects described on this site
Also working with
CH.06 · Services
Four things we do on every engagement.
Fixed and staged, priced in euros, one agreement. With a product or with the problem alone.
StrategyArchitectureGovernance
Consulting
decideWhich layer, which runtime, which team first, what stays local. You leave with an architecture, and a reason for each choice.
- AI Discovery SprintTwo weeks for a fixed fee. You get one prioritised backlog and a clear go or no-go.
- Solution architectureWe choose the model, the runtime and the connectors, and agree where your data may go and where it may not.
- Governance designWe set who reaches what, under which limits, and what gets recorded.
Machine learningGenerative AIAI Skin
Implementation
buildWe build on your systems and stay until it runs in production. One team covers machine learning, generative AI, agents and skins.
- Proof of valueA working agent or skin on your data in weeks, with the failure cases agreed before anyone trusts it.
- Machine learningChurn, forecasting, and federated learning for when the data cannot move.
- Generative AIRAG, agentic workflows, fine-tuning, local LLMs.
OperationsAtalaia
Support and maintenance
runModels change, APIs move and rules drift. A named engineer answers when that happens, and you keep sight of every system we run for you.
- Model and API changesWe absorb and test them before they reach your users.
- Atalaia on the estateWe report costs, permissions and drift to the people who decide.
- A named engineerOne person who knows your systems answers the phone.
DevelopersBusiness teamsLeadership
Training
teachYour people learn to run and extend what we built. The skin keeps growing after we leave.
- DevelopersYour developers learn to code with agents against your house standards.
- Business teamsYour business teams learn to work inside the assistant, with the skills the skin ships.
- LeadershipPositioning and governance workshops for the people who decide.
The founders
Six people. You meet all of them.
Everyone here builds. There is no account layer between you and the engineer doing the work.
Miguel Vicente
Founder & Managing Partner
Business StrategySalesDigital Marketing
Miguel Vicente is the Founder and Managing Partner of Dypsis AI, where he guides companies through AI-driven transformations to optimize Business Strategies, models and performance.
LinkedIn
Frederico Vicente
Co-founder and Partner, Head of AI
LLMsMachine LearningSoftware Architecture
AI researcher and entrepreneur with 5+ years of experience in machine learning and business transformation.
LinkedIn
Miguel Vicente Jr.
Co-founder & Head of Operations
Operations ManagementProduct ManagementProcess Optimization
Miguel Vicente Jr. is the Co-founder and Head of Operations at Dypsis AI, where he drives operational excellence and ensures seamless delivery of AI solutions to clients.
LinkedInTake something with you
Start
Bring us the problem nobody has cracked yet.
We are a small team of senior specialists. We pick the right model and the right layer, and we build the least machinery that does the job. You get a call with an engineer, not a sales deck.