~/samuel-tagliacozzo · a.k.a. artysan
I build systems that work with AI agents.
Freelance developer in Rome. I design and ship products with AI inside, automations, and the infrastructure that holds them up. Everything is here: projects, activity, and where to find me.
Samuel Tagliacozzo
AI & full-stack engineer · freelance
- BuildingAgents Multi towards 1.0
- ShippingSEO Ops first deploy
- WorkingExam simulator
- WorkingLead qualification
- WorkingCustom management systems
- PlayingDragon's Lair rules engine
- StudyingAI from the inside
- InfrastructureHomelab and OVH servers
Pinned
Projects
My working environment with AI agents: isolated accounts, a local console, MCP servers and a memory service shared by every Claude I use.
A web companion for D&D 5e played at a physical table: sheets, rolls, initiative and codex. An action-based rules engine, built to become self-hostable.
From a site's real SEO data to a few reasoned actions, approved by a person and measured before and after. My product, with two partners.
Multi-tenant SaaS that talks to leads on WhatsApp and by phone, qualifies them and hands them to the right salesperson. In production, built by me.
Generates and grades original exam papers across all four skills, speaking included, with the cost of every generation recorded. In production.
Maintaining and building on an n8n instance with over a hundred workflows: payments, invoicing, email and notifications for a network of companies.
What I do
Three things, done well
Agentic systems
Environments where multiple agents work reliably: memory, tools, guards, verification. The most complete case is Agents Multi.
Products with AI inside
From spec to production: lead qualification, exam simulators, SEO. With model costs measured before setting a price.
Integrations and infrastructure
WhatsApp Business, payments, n8n with over a hundred workflows. My own servers, deploys and migrations without data loss.
Got a project, or want to talk agents?
Drop me a couple of lines: I'll answer personally. If needed, we start with a call to understand the problem before the code.