Agentic systems · Automation · Product design

I bring AI into processes, where it actually matters.

I study how the team works and turn manual tasks and disconnected tools into integrated workflows, copilots and agents — from prototype to adoption. I support adoption with training built around company roles and real use cases.

I reply within one business day

Available for new B2B projects
Alessio Marcone presenting innovation to a business audience
Alessio MarconeProduct thinking. Real systems.

What I do

Process first. Then the system.

I start from how the team actually works, identify where time and information are lost, and build the most useful path: strategy, prototype, integration and adoption.

01

AI strategy, adoption and training

I map processes, data and constraints, define a practical roadmap and design training around company use cases. The team knows what to automate, how to use it and under whose responsibility.

Process audit · Roadmap · Training
02

Custom products and copilots

I design RAG assistants and internal tools that work with company data, cite their sources and fit existing workflows.

RAG · APIs · Internal tools
03

Operational automation and agents

I connect systems and repetitive steps into observable workflows, with human approval wherever judgement matters.

Agents · Workflows · Human in the loop
04

Governance and responsible use

I make AI uses, data, responsibilities and controls explicit. AI Act programmes are delivered through Upion, a venture in validation.

Inventory · Policy · Controls

Operating stack

Different tools. One system.

The stack changes with the process. I use these tools to move quickly from workflow maps to interfaces, integrations and systems ready for day-to-day work.

I choose the stack after the problem: fewer dependencies, more control.
Design

From process to prototype

Flows, interfaces and components that can be tested before investing in production.

FigmaDesign systemsPrototyping
Build

From prototype to software

Coding agents and version control to build, verify and iterate quickly.

CodexClaude CodeGitHub
Orchestration

Tools working together

Workflows and system connections, with human control at decision points.

n8nMCPAPIs & webhooks
Knowledge

Answers connected to sources

Data and documents become a searchable, traceable and maintainable knowledge base.

RAGEmbeddingsSource-backed skills

Selected work

Product design, evolved with AI.

Three B2B systems designed around real work, now reframed as concrete opportunities for automation and decision support.

01Internal project · Service operations prototype

Helply

Design systemProduct designAI automation
Solution

I designed the internal product end-to-end: design system, components, early prototype and initial AI automations for intake, triage, routing and resolution support.

Value

The prototype shows how email can become a structured workflow while keeping people in control of operational decisions.

Origin
Internal project
Users
Support team
Volume
10 tickets / week
Status
Prototype

My roleFrom brief to product: design system, components, early prototype and AI automations.

Open the Figma project
Helply dashboard for B2B support ticket management
02Fintech operations · Product strategy + AI

Factoring Operations

Product strategyWorkflow automationAI copilot
Solution

An operations dashboard and client portal unify the cycle. The proposed AI evolution can check completeness, priority and next actions.

Value

A clearer, more scalable process from document retrieval to internal review.

Read the full case study
Operations dashboard for factoring services
03Repeatable template · Agents

Market operator hub

Claude agentsTelegram APIWorkflow
Solution

One agentic workflow coordinates collection, drafting, approval and publishing.

Value

A repeatable model for operators with similar processes, while keeping human approval before publishing.

Explore the workflow
Agentic content and communications workflow for a market operator
04Agentic operating system

Upion agent team

A more consistent, verifiable editorial process that is ready to evolve.

Claude CodeSubagentSkillHook
05Knowledge system · Agentic curation

Skill For AI

Selection becomes clearer and more verifiable, with explicit criteria instead of a simple directory.

Agentic curationSource auditScoring
Visit project
06Open source · Reliable workflows

Agent Fieldbook

Reusable operating practices that remain connected to their original source.

ClaudeCodexSource-backed skills
Visit project
07Venture in validation

Upion

The proposition and delivery model are being validated with the market. Not an established company.

VercelSEOAssessment
Visit project

Method

From bottleneck to usable system.

Each phase reduces uncertainty around the problem, the solution and the value created.

  1. 01

    Analyse

    We map the workflow, people, data and constraints, then choose the right intervention point.

  2. 02

    Prototype

    We make the solution tangible and test it against real work before investing in production.

  3. 03

    Produce

    We connect systems and data, define controls and prepare the team for day-to-day use.

  4. 04

    Measure

    We observe adoption, quality and operational impact, then improve what creates value.

About

I design and deploy AI systems.

I’m Alessio Marcone. I combine product design, automation and software building to turn an operational problem into a system the team can understand, integrate and use.

Based in Turin, working remotely with companies in Italy and abroad.

I also build web and creative coding projects — portfolio available on request.

Based
Turin, Italy
Work
Italy + international
Focus
AI applied to processes

Let’s talk

Let’s start with a process that costs time today.

Tell me what happens today, where the workflow gets stuck and which outcome you need. I’ll reply with a concrete first point of view.

hello@alessiomarcone.it I reply within one business day.