Dario Nasti. IT
Supply chain · Fashion sourcing · Agentic AI

From Bangladesh to the store floor. AI agents that do the follow-up.

I help brands, manufacturers and retail chains buy better, produce on time and put stock where it sells. 25 years in buying, supply chain and retail. 35 stores on ML forecasting. Direct sourcing in Bangladesh.

Try the demos See the consulting work
AWS ML SpecialtyClaude Partner NetworkNaples · Italy
What I do

I work where fashion meets data.

I come from the operational side: buying, factory cycles, inspections, negotiations. On that base I build systems that take repetitive work off buyers, merchandisers and product offices.

Operations

Supply chain and organisation

Org charts, SOPs and job descriptions written from staff interviews. Review of supply, quality and logistics contracts with proposed redlines. EBITDA models by client line, material coding and bill-of-materials control.

SOPContractsEBITDABills of materials
Far East

Sourcing and production follow-up

Denim and knitwear production in Bangladesh: FOB and CMT negotiation, production planning, AQL inspections, and fixing fabric composition and finishing issues.

FOB / CMTAQLDenimKnitwear
Retail

Forecasting and allocation

Demand forecasting by store and item, seasonal buying books, initial allocation down to size, stock rebalancing between stores and markdown analysis of slow sellers.

LightGBMXGBoostDeepARAWS SageMaker
Agentic AI

Agents for product offices and buyers

Source Mind automates the product office: it reads supplier emails and WhatsApp messages, updates the tracker, flags delays and drafts replies. RetailMind automates the buyer, store planner and retail analyst roles. MCP servers connected to the company ERP.

MCPClaudePythonWorkflows
Digital

Marketplaces and e-commerce

Multi-country catalogues, GPSR and EPR compliance, and a Shopify-plus-marketplace strategy that also clears stock from past seasons.

GPSREPRShopify
Consulting

Real projects, confidential clients.

Client names stay private. The figures and systems are the ones actually delivered.

Luxury leather goods · Naples

ERP-connected order tracker for a manufacturer producing for international luxury houses

An MCP server on the company ERP plus a daily pipeline: Excel refresh, planning and GANTT, MRP, a 7-tab HTML tracker and an AI-written report. Alongside it: org chart, SOPs, an escalation procedure for late POs, review of framework supply and quality agreements, and EBITDA models by client line.

9MCP tools on the ERP
5daily pipeline steps
7tracker tabs
Menswear retail · 35 stores

ML forecasting and the seasonal buying book

  • AWS pipeline built from scratch: SageMaker, Lambda, API Gateway, React dashboard
  • Two-stage hurdle model: probability of sale, then quantity
  • Allocation down to store × item × colour × size
1.8Mrows, 2022–2026
~31.5%WAPE by cluster
171,756units in the buying book
Try the rebalancing demo →
Furniture · Manufacturing

Bill-of-materials and margin control

  • Prototype bill of materials and the process for loading it into the ERP
  • BI on margins, material consumption and variances
  • Second module: measuring production times, which are not tracked today
2modules: margins and times
Denim · Bangladesh → Italy

Sourcing agent for an Italian buyer

  • FOB negotiation across several collections: FW26, SS27, FW27
  • Production follow-up with the factory and quality control
3collections negotiated
Try the AQL calculator →
Wholesale fashion · Marketplace

One catalogue across several EU countries

  • Catalogue extended from Italy to France and Spain
  • GPSR compliance, EPR registrations for packaging and textiles
  • Minimum sustainable prices calculated on real cost
606active products
Try the minimum price calculator →
Demos

Try the work, not the promises.

Four tools that reproduce my consulting work. They run right here, on sample data.

Carmen, the product office agent

She reads supplier messages, works out which order they refer to, updates the tracker, alerts the buyer and drafts the reply.

Sample data
Incoming messages
Agent reasoning
Waiting for messages to process.
OrderSupplierStageEx-factoryStatus
News

What I'm working on now.

Source Mind tested on real emails

The agent was run on a real, anonymised archive of emails between a buyer and a denim factory. The result becomes the demo for upcoming clients.

Agentic AI · Source Mind

Furniture: production times module kicks off

After bills of materials and margins, the client asked to measure processing times, which are not tracked today. Project plan and quote in preparation.

Consulting · Manufacturing BI

Two agents, two roles

Source Mind builds agents and workflows for product offices. RetailMind does the same for buyers, store planners and retail analysts. Forecasting stays statistical; the innovation is orchestrating people through reports and alerts.

Startup

Fashion catalogue live in France and Spain

A wholesale catalogue extended from Italy to other EU countries, with the EPR registrations required for packaging and textile products.

Marketplace · Compliance

SS2027 buying book delivered

About 171,000 units planned for a 35-store network, with allocation down to size and corrected twin-item assignments.

Forecasting · Retail
Career

Twenty-five years inside fashion.

Today

Supply chain and AI consultant

Independent · co-founder of Source Mind and RetailMind

Luxury manufacturing, fashion retail, furniture and Bangladesh sourcing. Building AI agents for product offices and buyers.

2024 – 2026

Temporary Digital Supply Chain Manager

Camomilla Italia

Purchasing analysis, forecasting, KPIs and supply chain visibility. Introduced AI tools into the processes.

2024 – 2025

Chief Marketing Officer

PheasyAI
2020 – 2023

Supply Chain Director, then Sales & Stores Director

Doppelgänger Roma

Supply chain and goods-flow optimisation, then leadership of sales and the store network.

2001 – 2019

From Senior Retail Buyer to Deputy General Manager

Capri Srl · Alcott & Gutteridge

Buyer (2001–2003): OTB, replenishment, stock-to-sales ratio. Supply Chain Manager (2004–2010): production transitions and delivery models. Deputy General Manager (2010–2019): budgets, P&L, KPIs across own and franchise stores, a team of 20 retail analysts.

Certifications

  • AWS Certified Machine Learning – Specialty
  • Claude Certified Architect, in preparation
  • Member of the Claude Partner Network

Tools

  • Python, SQL, AWS SageMaker
  • Claude Code, MCP, agents and skills
  • Advanced Excel, React for dashboards

Languages

  • Italian, native
  • English, professional
  • Spanish, basic
Contact

Tell me where your process gets stuck.

Orders arriving late, stock sitting in the wrong store, margins that don't add up. We start with a half-hour call and your real data.

[email protected]LinkedIn · Dario Nasti ↗Naples, Italy · also working remotely