I make the technology decisions a business runs on.

Fractional CTO work for product teams that need senior technical decisions, and AI agents for companies with repeatable processes.

Matteo Gazzurelli · Fractional CTO and agentic AI consultant · 20+ years leading technology · Brescia, Italy

Automating a process?Move the slider for a first estimate of the hours you could get back.

Hours per week on manual reporting · 12

530 h back per year

Full estimate Assumption: 85% of those hours recovered, over 52 weeks. Measure current hours first; verify savings in a pilot.

Twenty-plus years of technical judgement

I'm Matteo Gazzurelli, fractional CTO and agentic AI consultant with MaGa Srl in Brescia. 20+ years and 250+ projects taught me which technology a business can run on. Today I decide where AI agents fit, then see them into production.

20+Years leading technology

250+Projects led

GDGBrescia, co-founder and organizer

7Talks, from CERN 2012 to DevFest Milano 2026

AI agents: Gemini Enterprise Agent PlatformClaude Managed AgentsGoogle ADKClaude Agent SDKClaude CodeGeminiGenkitMCPA2AA2UI
Knowledge and RAG, on premise too: RAG with local modelsQwenMistral AIDeepSeekopen-weight embedding modelspgvectorQdrantDocling
Python backends: PythonFastAPIPydanticSQLAlchemyPostgreSQLpytest
TypeScript backends: Node.jsTypeScriptFastifyExpress.jsCloud FunctionsVitest
Web and mobile: Next.jsReactTailwind CSSshadcn/uiFlutterThree.jsPlaywright
Cloud and delivery: FirebaseFirebase App HostingCloud RunDockerGitHub ActionsCloud BuildSecret ManagerAnsible

More about me

How I choose the technology

I choose the technology on the case: the systems already in place, the data available, who will maintain it.

Agent or workflow?

The first question in every project. A plain workflow covers many repetitive tasks; an agent earns its place when the work needs judgement across several steps. I answer it before any code is written.

How I decide

Google, Anthropic or something else

Gemini Enterprise Agent Platform and Claude Managed Agents are the managed platforms; Google ADK and Claude Agent SDK are the frameworks for building and owning the agent. Which pair fits depends on your systems and your team.

Google vs Anthropic, as of October 2026 Google ADK integration

My position

The winner changes with the case. I tell you which option fits yours, and why, before we start.

How I work

An engagement takes one of three shapes. Each one ends with something your team keeps.

Assessment

I look at the process or the product as it runs today, with the people who run it, and I say what an agent would change and what your systems allow.

You get a written plan, with a go or no-go for each use case.

Build, with your team

I design and build the first agents alongside your developers, with a person checking the steps where a mistake would cost. Your team learns the stack as we build.

You get agents in production and the people who can maintain them.

Advisory over time

As fractional CTO I stay on as the senior technical voice: architecture, hiring, releases, and the calls on AI systems that today need a senior in the room.

You get technical decisions made, written down and followed through.

Prices are agreed after the first conversation, once the scope is clear.

Selected work

Three engagements, anonymised. The sector, what the client needed, what changed.

Industrial software, regulated releases
Releases took months and stakeholders saw progress only at the end.
Read the case
Releases about every three or four weeks, reliable dates, progress visible along the way.
Education SaaS
The product grew faster than the team's technical decisions.
Read the case
Architecture and priority decisions, written down, stopped being reopened every month.
Manufacturing SME, Northern Italy
Office work still ran on email, spreadsheets and people's memory.
Read the case
The person who re-keyed the data now checks the agent's work.