Backend and AI for products that need to actually work.

I build APIs, integrations, automations, and AI features for teams that need practical, maintainable software connected to real tools.

Chatbot + CRM

Problem

50 requests a day to read, qualify, and route to the right team. All done manually.

Result

AI chatbot that qualifies, answers FAQs, and assigns leads to the right salesperson. 80% handled without intervention.

Document automation

Problem

Invoices, orders, and documents opened one by one, data copied and entered into the ERP.

Result

Pipeline that extracts data automatically, validates it, and syncs with the ERP. From 3 hours a day to 15 minutes of review.

Automated reports

Problem

Weekly report compiled by hand: collect data from 5 different sources, format, send.

Result

System that collects, processes, and generates the report automatically. Ready in your inbox every Monday at 8am.

Services

What I build.

01

Custom AI assistants

For support, internal search, and lead qualification.

Assistants connected to your data and tools: knowledge bases, CRMs, websites, WhatsApp, or internal portals. They answer, find information, and help complete real operations.

Knowledge base CRM WhatsApp
02

Operational automations

For removing repetitive steps and manual work.

Workflows that remove manual steps: data collection, AI classification, scraping, reports, notifications, and scheduled jobs.

Scraping Reports Scheduled jobs
03

System integrations

For connecting tools that currently stay separate.

Connections between tools that do not talk today: CRMs, management systems, external APIs, webhooks, and data synchronization.

APIs Webhooks Data sync

Stack

The tools I use every day.

A pragmatic set for building, integrating, and deploying solutions that stay maintainable.

Backend & product

.NET
PostgreSQL
React
TypeScript

AI & Automation

OpenAI
Claude
RAG
Temporal

Deploy & infrastructure

Docker
GitHub
Cloudflare

About

Matteo Vannucchi

Italy / remote

Computer science degree and master's in Artificial Intelligence from the University of Bologna. I specialized in automation because I keep seeing the same problem: manual processes full of errors, systems that don't talk to each other, people acting as glue between tools that should work on their own.

I respond quickly, keep meetings short, and prefer showing a working prototype over slides. If you have a process that depends on someone remembering to do it, we can automate it.

Process

A useful version first, then the rest.

I do not start with huge architectures or months of specs. I start from the real process, find a high-impact step, and build a first version you can test with real data and tools.

  1. 01

    Start from a real case

    We choose a concrete process: customer requests, documents, leads, reports, synchronizations, or repetitive tasks.

  2. 02

    Build something usable

    Not a fake demo: a first version connected to your data, your tools, or a real workflow.

  3. 03

    Measure whether it saves time

    We look at where it works, where it gets stuck, and what is worth automating better.

  4. 04

    Only then expand the system

    If the flow holds up, we make it more reliable: permissions, logs, edge cases, deploys, and integrations.

Contact

Let's talk about the problem you want to solve.

Whether it is a backend, an AI feature, an integration, or a process to automate, tell me what is not working today and what you would like to achieve.

What to include

  • Problem
  • How it works today
  • Expected outcome