Fractional Product Manager, AI automation and Product Builder in Milan

What I do

As a fractional product manager and AI builder, I help you move your digital product out of the build trap: do less and do it better. Whether you're a founder with an idea to build or a startup or scaleup with a product to grow, I put structure and processes in place so your team can make decisions more independently, and I bring AI-assisted development into its day-to-day work. When needed, I design and build products and automations myself using n8n, RAG and CAG, reducing repetitive work so the team has more time to focus on the product.

My approach

Before deciding what to build, I work with the team to clarify the problem, starting from the business outcomes we need to achieve. We then form solution hypotheses and validate them in the field, testing and combining data with direct conversations with the people who use the product every day. The roadmap takes shape from a sequence of evidence that guides priorities, so the team knows why it is investing and what it wants to learn from each release.

One example is Facile.it, Italy's largest consumer credit brokerage network: more than 300 people across physical stores, with applications still relying on paper, manual signatures and commissions calculated by hand. The risk wasn't just digitizing the process. It was building a product disconnected from the brokers' day-to-day work. That's why I worked alongside them, using interviews, incremental releases and validation demos, and measuring what worked at each step. The paperless workflow reached 80% adoption in two months and saved around €150K per year. Over the same period, conversion on consumer credit proposals increased by 25% compared with the previous year.

A few real results

Three different projects, but the responsibility stays the same: understand the problem, decide what's worth building and measure results at every iteration. I led each from idea to production, setting the product strategy and, when the project called for it, building the solution myself.

What I automate

I work on process automation, starting with a service blueprint assessment and one clear goal: free up the team's capacity while making repetitive, manual processes more reliable. Using n8n, Make or Zapier, I design and build complex workflows, knowledge bases and second-brain systems on RAG and CAG infrastructure, so company know-how is stored, kept up to date, searchable and reusable when needed. One example is a cross-brand n8n architecture for the post-purchase flows of four eCommerce sites: catalog feeds, carrier integration, third-party systems and individual service delivery. An internal dashboard completed the project, supporting governance and operational monitoring.

One example I like to share started shortly before my honeymoon: while I was away, I risked becoming a bottleneck for the team. So I built an AI version of myself on n8n. An ingestion flow pulled content from the repository containing the knowledge base I had documented, while a RAG workflow with a feedback loop made it queryable through Slack. The team could access my knowledge without depending on my availability and, by the time I returned, had already used the workflow more than 80 times. That project brought me closer to the n8n community, where I advocate for using these tools more thoughtfully inside companies.

What I build

With AI coding, generating code isn't the problem. The problem is understanding and reviewing decisions that accumulate faster than the team can keep up with. Tierward grew out of the experience and repeated failures of my first projects. It's an open-source framework that adds a solid, governed harness to the product development process. Today it is built around Claude Code, with Codex on the roadmap. Through a phased process and mechanical gates, it orchestrates the functions of a product team, from scoping through quality assurance, and guides development end to end. The goal is to put a harness around AI, place it inside a flow the team can control and keep the decisions visible and understood.

Tierward automatically detects 11 tech stacks and lets you choose how much structure to apply: from Fast Lane for contained changes to the Full pipeline for more complex features and team-based work. The process scales with the risk and the level of governance required, without placing the same overhead on every task. The principle is simple: AI generates, but control stays with you or your team.