AI Agents for Supply Chain Optimisation with n8n

From production planning to transport management and CO₂ estimation, discover three agentic workflows to help you optimise your supply chain.

Need Help?
Subscribed! Error
AI Agents for Supply Chain Optimisation with n8n

As a supply chain engineer and data scientist, I’m convinced advanced analytics can solve many day-to-day operational problems.

What is the bottleneck? Most of the time, it is the adoption by users.

Supply Chain Planners: “Can we receive the optimization tool outputs by email?”

In User Acceptance Tests (UAT), users consistently request results that integrate with their existing workflows (email, Excel, PowerPoint).

Our solution involves embedding supply chain optimisation analytics into existing workflows using n8n.

The agent wrapper drawn as four steps with what passes between them boxed underneath: an email with a body and an attach
Example of the type of workflow we will introduce in this article — (Image by Samir Saci)

We can connect algorithms, wrapped in API microservices, to user workflows through AI agents equipped with HTTP tools.

The agent reads the request, calls the proper backend and replies with an enriched answer.

In this article, I will introduce three agentic workflows built with n8n:

  • A Production Planning Agent that provides the optimal production plan based on orders received by email.
  • A Transport Management Agent that computes the optimal route to deliver multiple stores and sends a confirmation by email to customers.
  • A Freight CO₂ Emissions Agent that structures shipment data and queries the Carbon Interface API to provide a footprint summary.

These supply chain optimisation workflows can be deployed in your own instance following the tutorials linked in this article.

AI Workflows Connecting Emails to APIs

The idea is to create real value for supply chain operations while integrating seamlessly into existing processes.

To illustrate this approach, we will use the example of a production planning optimisation algorithm, built in Python, that we transformed into an Agentic workflow using FastAPI and n8n.

Production Planning Optimisation Module

Let us imagine you received a customer order with quantities to be delivered over the next 12 months.

Matplotlib bar chart of the demand forecast used as the input of the production planning model, twelve periods with a di
Example of Customer Order — (Image by Samir Saci)

The objective of the planning team is to find the optimal plan to minimise production costs, considering:

  • Setup Costs: fixed costs you have each time you set up a production line
    Example: 500 $ per production batch
  • Holding Costs: cost of storage per unit per time
    Example: 1 $/unit/month

If you produce the exact quantity needed, you will minimise holding costs, but setup costs will explode (12 setups).

Plotly grouped bar chart of the lot for lot plan, forecast in blue and production in red identical in every one of the t
Example of high setup costs plan — (Image by Samir Saci)

On the contrary, if you produce the total quantity in a single batch, you will minimise the setup costs (only one setup), but holding costs will explode.

The opposite extreme of a16-02: one single production batch of 2,000 units in period 1 towering over the twelve forecast
Single batch plan — (Image by Author)
How to find the right balance?

In another article, I explain how I use the Wagner Within algorithm with Python to find the optimal production schedule.

The optimal plan the solver returns, four production batches covering the same 2,000 units of demand: 550 in period 1, 4
Optimal Production Plan — (Image by Author)

This solution has been packaged in a FastAPI microservice with multiple endpoints

  • /upload_prod: this endpoint receives a POST request with the demand dataset included to upload it to the backend
  • /launch_plan: this endpoint receives a GET request with parameters like setup cost, holding cost, and time unit
The FastAPI backend behind the production planning agent, two endpoints and the request type each expects. It is deliber
FastAPI Backend that can be connected to n8n workflows — (Image by Samir Saci)

Initially, we aimed to integrate this endpoint with the front end of a web application, where users can upload demand data and select parameters.

Full screenshot of the LogiGreen App production planning module, left navigation and parameter form on the left, results
Example of UI of the LogiGreen Apps Demo — (Image by Samir Saci)

However, production planners wanted something more integrated into their current workflow.

As they received many requests for quotations from the commercial teams by email, they wanted to have the solution connected to a specific mailbox.

The target workflow in one line: a customer order arriving by email on the left, an agent box holding an API tile and an
Requested Workflow by Production Planners — (Image by Author)

The idea would be to have an AI workflow that extracts the request from the email (including attachment and body), runs the algorithm, and replies with a detailed quote.

We can prototype this with the support of n8n!

A Simple Architecture in n8n

Planners receive requests via email that include details in the body and requested volumes by period in the attachment.

Gmail screenshot of the request that triggers the whole workflow, a plain business email asking for a production plan an
Example of Email received by the planning team (Image by Samir Saci)

It has been agreed with the commercial team that they will follow a specific format of emails:

  • Attachment: demand dataset in (.csv) format
  • Email body: production planning parameters like holding costs, setup costs, and unit of measure
The incoming email the parser has to read, annotated with two green boxes and slider icons naming what the agent extract
Example of Request Received by Email — (Image by Samir Saci)

This will be the input of our AI workflow created with n8n.

The n8n canvas for the production planning workflow, with green execution ticks and item counts on the connectors, so it
AI Workflow for fully automated email reply — (Image by Samir Saci)

Step 1: Collect Email and Download the Attachment

The GMAIL trigger node collects the email body and downloads the attachment.

The first three nodes of the n8n workflow, Gmail Trigger to Extract CSV from Attachment to POST Upload CSV, with the ite
Focus on the upload of demand data — (Image by Samir Saci)

The (.csv) file is converted into JSON and sent via POST request to the backend.

Now that the dataset is uploaded, we can provide the email body to the AI Agent Parser.

Step 2: Collecting the Production Plan Parameters

The AI Agent Parser parses the email content to extract the parameters, which are returned in JSON format.

The reasoning path of the same workflow: Gmail Trigger into an AI Agent Parser, then an AI Agent API Request, then a Gma
Focus on the AI Agent Parser — (Image by Samir Saci)

The system prompt details how to parse the email to collect the proper parameters.

The parser system prompt as published, a bordered white card holding the whole instruction: role, the statement that the
System Prompt of the AI Agent Parser — (Image by Samir Saci)

The outputs of this first agent are sent to the AI Agent equipped with the API query tool, which is connected to the FastAPI microservice.

The n8n output panel showing the five fields the parser returned from the email in a16-08, and every value matches the e
Outputs of the AI Agent Parser from the email shared above — (Image by Samir Saci)

I use a minimal system prompt to show this second agent how to use the API we have deployed.

I provide an overview of the parameters available:

The second system prompt of the article, the one given to the agent that reads the model results rather than the email.
Overview of parameters in the system prompt — (Image by Samir Saci)

I don’t forget to list the outputs of the API:

The output specification part of the analyst prompt, listing the model results the agent is allowed to talk about. The l
List of the outputs in the system prompt — (Image by Samir Saci)

And I finish with the task expected:

The task block of the analyst prompt, the part that turns numbers into a business summary. It is the most instructive of
Example of Summary Provided by the AI Agent — (Image by Samir Saci)

The output is sent back to the commercial manager via email using the last Gmail node.

A production planning result delivered as an ordinary Gmail message from the LogiGreen bot, with demand, the months chos
Example of Summary Provided by the AI Agent — (Image by Samir Saci)

The summary includes a reminder of the input parameters, a detailed description of the production plan, and the overall cost structure that will be used for the quote.

👉 Check the video linked below for a live demo of the workflow

As a supply chain engineer and data scientist, I’m convinced advanced analytics can solve many day-to-day operational problems.

What is the bottleneck? Most of the time, it is the adoption by users.

Supply Chain Planners: “Can we receive the optimization tool outputs by email?”

In User Acceptance Tests (UAT), users consistently request results that integrate with their existing workflows (email, Excel, PowerPoint).

Our solution involves embedding supply chain optimisation analytics into existing workflows using n8n.

The agent wrapper drawn as four steps with what passes between them boxed underneath: an email with a body and an attach
Example of the type of workflow we will introduce in this article — (Image by Samir Saci)

We can connect algorithms, wrapped in API microservices, to user workflows through AI agents equipped with HTTP tools.

The agent reads the request, calls the proper backend and replies with an enriched answer.

In this article, I will introduce three agentic workflows built with n8n:

  • A Production Planning Agent that provides the optimal production plan based on orders received by email.
  • A Transport Management Agent that computes the optimal route to deliver multiple stores and sends a confirmation by email to customers.
  • A Freight CO₂ Emissions Agent that structures shipment data and queries the Carbon Interface API to provide a footprint summary.

These supply chain optimisation workflows can be deployed in your own instance following the tutorials linked in this article.

AI Workflows Connecting Emails to APIs

The idea is to create real value for supply chain operations while integrating seamlessly into existing processes.

To illustrate this approach, we will use the example of a production planning optimisation algorithm, built in Python, that we transformed into an Agentic workflow using FastAPI and n8n.

Production Planning Optimisation Module

Let us imagine you received a customer order with quantities to be delivered over the next 12 months.

Matplotlib bar chart of the demand forecast used as the input of the production planning model, twelve periods with a di
Example of Customer Order — (Image by Samir Saci)

The objective of the planning team is to find the optimal plan to minimise production costs, considering:

  • Setup Costs: fixed costs you have each time you set up a production line
    Example: 500 $ per production batch
  • Holding Costs: cost of storage per unit per time
    Example: 1 $/unit/month

If you produce the exact quantity needed, you will minimise holding costs, but setup costs will explode (12 setups).

Plotly grouped bar chart of the lot for lot plan, forecast in blue and production in red identical in every one of the t
Example of high setup costs plan — (Image by Samir Saci)

On the contrary, if you produce the total quantity in a single batch, you will minimise the setup costs (only one setup), but holding costs will explode.

The opposite extreme of a16-02: one single production batch of 2,000 units in period 1 towering over the twelve forecast
Single batch plan — (Image by Author)
How to find the right balance?

In another article, I explain how I use the Wagner Within algorithm with Python to find the optimal production schedule.

The optimal plan the solver returns, four production batches covering the same 2,000 units of demand: 550 in period 1, 4
Optimal Production Plan — (Image by Author)

This solution has been packaged in a FastAPI microservice with multiple endpoints

  • /upload_prod: this endpoint receives a POST request with the demand dataset included to upload it to the backend
  • /launch_plan: this endpoint receives a GET request with parameters like setup cost, holding cost, and time unit
The FastAPI backend behind the production planning agent, two endpoints and the request type each expects. It is deliber
FastAPI Backend that can be connected to n8n workflows — (Image by Samir Saci)

Initially, we aimed to integrate this endpoint with the front end of a web application, where users can upload demand data and select parameters.

Full screenshot of the LogiGreen App production planning module, left navigation and parameter form on the left, results
Example of UI of the LogiGreen Apps Demo 

However, production planners wanted something more integrated into their current workflow.

As they received many requests for quotations from the commercial teams by email, they wanted to have the solution connected to a specific mailbox.

The target workflow in one line: a customer order arriving by email on the left, an agent box holding an API tile and an
Requested Workflow by Production Planners — (Image by Author)

The idea would be to have an AI workflow that extracts the request from the email (including attachment and body), runs the algorithm, and replies with a detailed quote.

We can prototype this with the support of n8n!

A Simple Architecture in n8n

Planners receive requests via email that include details in the body and requested volumes by period in the attachment.

Gmail screenshot of the request that triggers the whole workflow, a plain business email asking for a production plan an
Example of Email received by the planning team (Image by Samir Saci)

It has been agreed with the commercial team that they will follow a specific format of emails:

  • Attachment: demand dataset in (.csv) format
  • Email body: production planning parameters like holding costs, setup costs, and unit of measure
The incoming email the parser has to read, annotated with two green boxes and slider icons naming what the agent extract
Example of Request Received by Email — (Image by Samir Saci)

This will be the input of our AI workflow created with n8n.

The n8n canvas for the production planning workflow, with green execution ticks and item counts on the connectors, so it
AI Workflow for fully automated email reply — (Image by Samir Saci)

Step 1: Collect Email and Download the Attachment

The GMAIL trigger node collects the email body and downloads the attachment.

The first three nodes of the n8n workflow, Gmail Trigger to Extract CSV from Attachment to POST Upload CSV, with the ite
Focus on the upload of demand data — (Image by Samir Saci)

The (.csv) file is converted into JSON and sent via POST request to the backend.

Now that the dataset is uploaded, we can provide the email body to the AI Agent Parser.

Step 2: Collecting the Production Plan Parameters

The AI Agent Parser parses the email content to extract the parameters, which are returned in JSON format.

The reasoning path of the same workflow: Gmail Trigger into an AI Agent Parser, then an AI Agent API Request, then a Gma
Focus on the AI Agent Parser — (Image by Samir Saci)

The system prompt details how to parse the email to collect the proper parameters.

The parser system prompt as published, a bordered white card holding the whole instruction: role, the statement that the
System Prompt of the AI Agent Parser — (Image by Samir Saci)

The outputs of this first agent are sent to the AI Agent equipped with the API query tool, which is connected to the FastAPI microservice.

The n8n output panel showing the five fields the parser returned from the email in a16-08, and every value matches the e
Outputs of the AI Agent Parser from the email shared above — (Image by Samir Saci)

use a minimal system prompt to show this second agent how to use the API we have deployed.

I provide an overview of the parameters available:

The second system prompt of the article, the one given to the agent that reads the model results rather than the email.
Overview of parameters in the system prompt — (Image by Samir Saci)

I don’t forget to list the outputs of the API:

The output specification part of the analyst prompt, listing the model results the agent is allowed to talk about. The l
List of the outputs in the system prompt — (Image by Samir Saci)

And I finish with the task expected:

The task block of the analyst prompt, the part that turns numbers into a business summary. It is the most instructive of
Example of Summary Provided by the AI Agent — (Image by Samir Saci)

The output is sent back to the commercial manager via email using the last Gmail node.

A production planning result delivered as an ordinary Gmail message from the LogiGreen bot, with demand, the months chos
Example of Summary Provided by the AI Agent — (Image by Samir Saci)

The summary includes a reminder of the input parameters, a detailed description of the production plan, and the overall cost structure that will be used for the quote.

A Proof of Concept Validated

This simple workflow has been deployed for the last 12 weeks with a nearly flawless execution.

Using n8n to prototype it helped us to understand how LLM can interact with complex optimisation algorithms quickly.

In the next section, I will detail how we replicated the same framework for external APIs, providing us with access to infinite optimisation solutions.

AI Agents for Transportation Planning

As the first project involved a FastAPI microservice developed by us, we wanted to experiment with the use of agentic workflows with external APIs.

For a transportation company based in the Netherlands, we first implemented a simple (non-agentic) workflow to calculate driving times and distances using the Open Route Service API.

Could we improve the user experience by adding an AI layer on top?

Later, we decided to add two AI agents to connect them to the mailbox of the admin teams that manage shipment requests.

What this agent solves

The primary challenge for small and medium-sized transportation companies is that they still receive most requests via email, as shown below.

The pickup request email that drives the transport agent, a nine field shipment brief written as prose bullets by a name
Example of Pickup Request Received from a Retailer — (Image by Samir Saci)

Admin teams must copy and paste addresses into maps, check drive times, and send confirmations back to customers.

These parameters will also need to be manually entered into the system for recording.

How can we automate this?

This agent turns a customer email into a validated route and a professional confirmation reply, while logging every detail to Google Sheets for traceability.

How does it work?

Admin teams receive pickup requests via email (as shown in the example above) that include details in the body.

This will be the input of our AI workflow, created with n8n, that includes two AI Agent nodes:

  • AI Agent Parser is parsing the email to extract the shipment information (pickup location, delivery location, pickup time …)
  • AI Agent Reply that uses the shipment parameters along with the distance and time to send shipment confirmation

Between these two agents, you have a set of nodes to query the Open Route Service API to collect and record distances, GPS coordinates and driving time.

The whole transport agent workflow on one canvas, and the shape is the message: a single AI Agent Parser fans out into f
AI Workflow for fully automated email reply — (Image by Samir Saci)
  1. Gmail Trigger captures a new shipment request.
  2. AI Agent (Parser) extracts structured fields (pickup/delivery, time windows, temperature control, contact).
  3. Google Sheets stores the request.
  4. The Open Route Service geocodes addresses, then computes driving distance and ETA (using the HGV profile).
  5. AI Agent Reply drafts a confirmation that repeats the key details.
  6. Gmail sends the confirmation back to the requester.

Step 1: Parse the parameters from the Email

The GMAIL trigger node collects the email body that is sent to the AI Agent Parser.

A zoom on the left half of a16-39, the fan out from the AI Agent Parser into the four collector nodes and on to the thre
Focus on the email parsing agent — (Image by Samir Saci)

We use the system prompt to instruct the agent on how to parse the email to extract the shipment information.

The transport parser system prompt in a navy bordered card, defining the role and the eleven field JSON schema the agent
System Prompt of the AI Agent Parser — (Image by Samir Saci)

We can then collect, in JSON format, all the parameters included in the email.

The n8n schema view of what the parser returned from the email in a16-38, eleven string fields with every value matching
Outputs of the AI Agent Parser — (Image by Samir Saci)

Step 2: Querying the Open Route Service API

These parameters are recorded in a Google Sheet.

We are then using the HTTP nodes to query the Open Route Service API to collect the GPS coordinates of the pickup and delivery locations.

The geocoding HTTP node opened up, parameters on the left and the OpenRouteService response on the right. It is the one
Outputs of the Geocoding HTTP Query Nodes — (Image by Samir Saci)

These coordinates are then used to calculate distances with the routing function of the API.

The outputs are all recorded in the Google Sheet.

The enriched shipment record as a plain two column table, the eleven parsed fields plus the four the APIs added: pickup
Outputs of the pickup requests with distance and driving time — (Image by Samir Saci)

Step 3: AI Agent Reply send the shipment confirmation

We send these outputs to the AI Agent Reply, which we instruct to prepare a shipment confirmation in a specific format.

The system prompt of the reply agent, which is a writing brief rather than an extraction brief: tone, the instruction to
System Prompt of the AI Agent Reply — (Image by Samir Saci)

The output is sent back to the sender using the Gmail node.

The confirmation email the reply agent produced, addressed back to the coordinator who sent a16-38 and repeating ten shi
Pickup Confirmation — (Image by Samir Saci)

The shipment confirmation includes a summary of the pickup information, with the driving distance and time estimated with the API.

👉 Check the video linked below for a live demo of the workflow (with the template included in the description)

A Proof of Concept Validated

This simple workflow has been deployed for the last 12 weeks with a nearly flawless execution.

Using n8n to prototype it helped us to understand how LLM can interact with complex optimisation algorithms quickly.

In the next section, I will detail how we replicated the same framework for external APIs, providing us with access to infinite optimisation solutions.

AI Agents for Transportation Planning

As the first project involved a FastAPI microservice developed by us, we wanted to experiment with the use of agentic workflows with external APIs.

For a transportation company based in the Netherlands, we first implemented a simple (non-agentic) workflow to calculate driving times and distances using the Open Route Service API.

Could we improve the user experience by adding an AI layer on top?

Later, we decided to add two AI agents to connect them to the mailbox of the admin teams that manage shipment requests.

What this agent solves

The primary challenge for small and medium-sized transportation companies is that they still receive most requests via email, as shown below.

The pickup request email that drives the transport agent, a nine field shipment brief written as prose bullets by a name
Example of Pickup Request Received from a Retailer — (Image by Samir Saci)

Admin teams must copy and paste addresses into maps, check drive times, and send confirmations back to customers.

These parameters will also need to be manually entered into the system for recording.

How can we automate this?

This agent turns a customer email into a validated route and a professional confirmation reply, while logging every detail to Google Sheets for traceability.

How does it work?

Admin teams receive pickup requests via email (as shown in the example above) that include details in the body.

This will be the input of our AI workflow, created with n8n, that includes two AI Agent nodes:

  • AI Agent Parser is parsing the email to extract the shipment information (pickup location, delivery location, pickup time …)
  • AI Agent Reply that uses the shipment parameters along with the distance and time to send shipment confirmation

Between these two agents, you have a set of nodes to query the Open Route Service API to collect and record distances, GPS coordinates and driving time.

The whole transport agent workflow on one canvas, and the shape is the message: a single AI Agent Parser fans out into f
  1. Gmail Trigger captures a new shipment request.
  2. AI Agent (Parser) extracts structured fields (pickup/delivery, time windows, temperature control, contact).
  3. Google Sheets stores the request.
  4. The Open Route Service geocodes addresses, then computes driving distance and ETA (using the HGV profile).
  5. AI Agent Reply drafts a confirmation that repeats the key details.
  6. Gmail sends the confirmation back to the requester.

Step 1: Parse the parameters from the Email

The GMAIL trigger node collects the email body that is sent to the AI Agent Parser.

A zoom on the left half of a16-39, the fan out from the AI Agent Parser into the four collector nodes and on to the thre

We use the system prompt to instruct the agent on how to parse the email to extract the shipment information.

The transport parser system prompt in a navy bordered card, defining the role and the eleven field JSON schema the agent

We can then collect, in JSON format, all the parameters included in the email.

The n8n schema view of what the parser returned from the email in a16-38, eleven string fields with every value matching

Step 2: Querying the Open Route Service API

These parameters are recorded in a Google Sheet.

We are then using the HTTP nodes to query the Open Route Service API to collect the GPS coordinates of the pickup and delivery locations.

The geocoding HTTP node opened up, parameters on the left and the OpenRouteService response on the right. It is the one

These coordinates are then used to calculate distances with the routing function of the API.

The outputs are all recorded in the Google Sheet.

The enriched shipment record as a plain two column table, the eleven parsed fields plus the four the APIs added: pickup

Step 3: AI Agent Reply send the shipment confirmation

We send these outputs to the AI Agent Reply, which we instruct to prepare a shipment confirmation in a specific format.

The system prompt of the reply agent, which is a writing brief rather than an extraction brief: tone, the instruction to

The output is sent back to the sender using the Gmail node.

The confirmation email the reply agent produced, addressed back to the coordinator who sent a16-38 and repeating ten shi

The shipment confirmation includes a summary of the pickup information, with the driving distance and time estimated with the API.

👉 Check the video linked below for a live demo of the workflow (with the template included in the description)

AI Agents for Supply Chain Sustainability

from

CO2 reporting remains a primary challenge for small and medium-sized companies, as they struggle to accurately quantify transportation-related emissions.

Let us assume you are working for the logistics department of a retail company in France.

The second version of the confirmation email, wrongly flagged as a duplicate of a14-11 by the hash. It is the same shipm

You will receive a pickup confirmation similar to the one above, which includes the pickup location, the expected pickup time, and the quantity of shipments.

Your job is to:

  • Estimate the CO2 emissions of the shipment
  • Record these parameters in the system
How can we automate that using AI agents?

How does it work?

A shorter variant of the workflow aimed at emissions rather than routing: Gmail Trigger, AI Agent Parser, Record Shipmen
  1. Gmail Trigger captures a shipment email.
  2. AI Agent Node parses the email into strict JSON (addresses, times, distance, weight, etc.).
  3. Google Sheets records the shipment metadata (keyed by shipment_number).
  4. HTTP requests call the Carbon Interface API to estimate CO₂.
  5. Google Sheets updates the same row with carbon_kg and estimated_at (timestamp).

In this AI workflow, we utilise only the AI Agent Parser, which collects the parameters from the email.

Nearly identical in layout to a16-42, which is why the hash merged them, but it is a different record: the times are ISO

These parameters are used to query the Carbon Interface API, which returns the emissions.

The carbon estimate returned for the same shipment, an n8n schema view of a third party emissions API response. The inte

The remaining nodes will store the data directly in Google Sheets, which are connected to CO2 emissions reporting tools.

You can implement this workflow on your instance using the template shared in the link below.

🚚 CO2 Emissions of Freight Shipments with Carbon Interface API and GPT-4o | n8n workflow template
Tags: Sustainability, Supply Chain, AI Agent, CO2 Emissions, Carbon Interface API, Logistics, Automation* Context Hi…
n8n.io

For more templates of workflow automation for sustainability, have a look at this tutorial.

Conclusion

💡
If you have any question, feel free to ask here: Ask Your Question

Agentic Workflow as a new user interface

Working on these prototypes allowed us to experiment with a new stage in the productisation of analytics products with AI Agents.

Users: Can you help us maximize the metric XXX while respecting constraints YYY?

Everything starts with an operational problem that could be solved with optimisation or advanced analytics.

Two-panel setup for a network design problem, demand on the left as store icons across a world map with three input tile

You start to draft a solution in a Jupyter Notebook, similar to the one shared in this article.

As users cannot run Python on their machines, you want to deploy these in web applications, such as the one we built for Production Planning, presented in the video below.

Finally, you can reuse the FastAPI backend to include this solution in any workflow using LangGraph (and LangChain) or n8n.

The FastAPI backend behind the production planning agent, two endpoints and the request type each expects. It is deliber

For example, we have deployed AI workflows connected to

  • Excel and Google Spreadsheets to automate the root cause analysis of logistic performance issues
  • Teams and Slack chatbot to help non-technical users use optimisation engines with natural language 
    (e.g.: What would the production cost be if we reduce the volumes by 25%?)

About Me

Let’s connect on Linkedin and Twitter; I am a Supply Chain Engineer using data analytics to improve Logistic operations and reduce costs.

For consulting or advice on analytics and sustainable Supply Chain transformation, feel free to contact m.

Need Help?

Read next