social media, somebody claims their “AI agent” will run your complete enterprise when you sleep.
It’s as if they’ll deploy AGI throughout factories, finance groups, and customer support utilizing their “secret” n8n template.
My actuality examine is that many corporations are nonetheless struggling to gather and harmonise knowledge to observe fundamental efficiency metrics.
Logistics Director: “I don’t even know what number of orders have been delivered late, what do you suppose your AI agent can do?”
And these marketed AI workflows, which are sometimes not prepared for manufacturing, can sadly do nothing to assist with that.
Due to this fact, I undertake a extra pragmatic method for our provide chain initiatives.
As an alternative of promising an AGI that may run your complete logistics operations, allow us to begin with native points hurting a particular course of.
Logistics Director: “I need our operators to eliminate papers and pens for order preparation and stock cycle depend.”
More often than not, it entails knowledge extraction, repetitive knowledge entry, and heavy admin work utilizing handbook processes which are inefficient and lack traceability.
For instance, a buyer was utilizing paper-based processes to organise stock cycle counts in its warehouse.
Think about a list controller who prints an Excel file itemizing the places to examine.
Then he walks via the alleys and manually data the variety of bins at every location on a kind just like the one beneath.
At every location, the operator should pause to document the precise amount and ensure that the world has been checked.
We are able to (and should) digitalize this course of simply!
That is what we did with a Telegram Bot utilizing n8n, linked to a GPT-powered agent, enabling voice instructions.
Our operator now solely must observe the bot’s directions and use audio messages to report the variety of bins counted at every location.
This native digitalisation turns into the primary concrete step within the digital transformation of this low-data-maturity firm.
We even added logging to enhance the traceability of the method and report productivities.
On this article, I’ll use two real-world operational examples to point out how n8n can assist SMEs’ provide chain digital transformations.
The concept is to make use of this automation platform to implement easy AI workflows which have an actual impression on operations.
For every instance, I’ll present a hyperlink to a whole tutorial (with a GitHub repository containing a template) that explains intimately learn how to deploy the answer in your occasion.
Vocalisation of Processes
In logistics and provide chain operations, it’s all the time about productiveness and effectivity.
Provide Chain Answer Designers analyse processes to estimate the optimum productiveness by analysing every step of a activity.
A breakthrough was the implementation of voice-picking, additionally referred to as vocalisation.
The concept is to have the operators talk with the system by voice to obtain directions and supply suggestions with interactions like this one:
- Voice Selecting: “Please go to location A, choose 5 bins.”
- Operator: “Location A, 5 bins picked.”
- Voice Selecting: “Please go to location D, choose six bins.”
- Operator: “Location D, six bins picked.”
This boosts operators’ productiveness, as they now want solely deal with selecting the proper portions on the correct places.
However these options, usually supplied by Warehouse Administration System distributors, could also be too costly for small operations.
That is the place we will use n8n to construct a light-weight resolution powered by multimodal generative AI.
Vocalisation of Stock Cycle Rely
I need to come again to the preliminary instance to point out you the way I used Textual content-To-Speech (TTS) to digitalise a paper-based course of.
We assist the inventory administration staff at a medium-sized style retail warehouse.
Repeatedly, they conduct what we name stock cycle counts:
- They randomly choose storage places within the warehouse
- They extract from the system the stock stage in bins
- They examine on the location the precise amount
For that, they use a spreadsheet like this one.
Their present course of is very inefficient as a result of the inventory counter should manually enter the precise amount.
We are able to substitute printed sheets with smartphones utilizing Telegram bots orchestrated by n8n.
The operator begins by connecting to the bot and initiating the method with the /begin command.
Our bot will take the primary unchecked location and instruct the operator to go there.
The operator arrives on the location, counts the variety of bins, and points a vocal command to report the amount.
The amount is recorded, and the situation is marked as checked.
The bot will then routinely ask the operator to maneuver to the following unchecked location.
If the operator’s vocal suggestions accommodates an error, the bot asks for a correction.
The method continues till the ultimate location is reached.
The cycle depend is accomplished with out utilizing any paper!
This light-weight resolution has been carried out for 10 operators with cycle counts orchestrated utilizing a easy spreadsheet.
How did we obtain that?
Allow us to take a look on the workflow intimately.
Vocalise Logistics Processes with n8n
A majority of the nodes are used for the orchestration of the completely different steps of the cycle depend.
First, now we have the nodes to generate the directions:
- (1) is triggering the workflow when an operator sends a message or an audio
- (6) guides the operator if he asks for assist or makes use of the mistaken command
- (7) and (8) are trying on the spreadsheet to seek out the following location to examine
For that, we don’t have to retailer state variables because the logic is dealt with by the spreadsheet with “X” and “V” within the checked column.
The important thing half on this workflow is within the inexperienced sticker
The vocalisation is dealt with right here as we accumulate the audio file within the Gather Audio node.
This file is distributed to OpenAI’s Audio Transcription Node in n8n, which offers a written transcription of our operator’s vocal command.
As we can’t assure that each one operators will observe the message format, we use this OpenAI Agent Node to extract the situation and amount from the transcription.
[SYSTEM PROMPT]
Extract the storage location code and the counted amount from
this quick warehouse transcript (EN/FR).
Return ONLY this JSON:
{"location_id": "...", "amount": "0"}
- location_id: string or null (location code, e.g. "A-01-03", "B2")
- amount: string or null (convert phrases to numbers, e.g. "twenty seven" → 27)
If a worth is lacking or unclear, set it to null.
No additional textual content, no explanations.[
{
"output": {
"location_id": "A14",
"quantity": "10"
}
}
]Due to the Structured Output Parser, we get a sound JSON with the required data.
This output is then utilized by the blocks (4) and 5)
- (4) will ask the operator to repeat if there may be an error within the transcription
- (5) is updating the spreadsheet with the amount knowledgeable by the operator if places and portions are legitimate
We now have now coated all potential situations with a strong AI-powered resolution.
Vocalisation of processes utilizing TTS
With this straightforward workflow, we improved inventory counters’ productiveness, decreased errors, and added logging capabilities.
We’re not promoting AGI with this resolution.
We remedy a easy drawback with an method that leverages the Textual content-To-Speech capabilities of generative AI fashions.
For extra particulars about this resolution (and how one can implement it), you may take a look at this tutorial (+ workflow)
What about picture processing?
Within the following instance, we are going to discover learn how to use LLMs’ image-processing capabilities to assist receiving processes.
Automate Warehouse Harm Reporting
In a warehouse, receiving broken items can shortly turn out to be a nightmare.
As a result of receiving can turn out to be a bottleneck to your distribution staff, inbound operations groups are below important strain.
They should obtain as many bins as doable so the stock is up to date within the system and shops can place orders.
After they obtain broken items, the entire machine has to cease to observe a particular course of:
- Damages must be reported with detailed data
- Operators want to connect footage of the broken items
For operators which have excessive productiveness targets (bins obtained per hour), this administrative cost can shortly turn out to be unmanageable.
Hopefully, we will use the pc imaginative and prescient capabilities of generative AI fashions to facilitate the method.
Inbound Harm Report Course of
Allow us to think about you might be an operator on the inbound staff on the similar style retail firm.
You obtained this broken pallet.
You’re supposed to organize a report that you just ship by e-mail, with:
- Harm Abstract: a one-sentence abstract of the problems to report
- Noticed Harm: particulars of the injury with location and outline
- Severity (Superficial, Average, Extreme)
- Really useful actions: return the products or fast fixes
- Pallet Info: SKU or Bar Code quantity
Fortuitously, your staff gave you entry to a newly deployed Telegram Bot.
You provoke the dialog with a /begin command.
You observe the directions and begin by importing the image of the broken pallet.
The bot then asks you to add the barcode.
Just a few seconds later, you obtain this notification.
Now you can switch the pallet to the staging space.
What occurred?
The automated workflow generated this e-mail that was despatched to you and the standard staff.
The report consists of:
- Pallet ID
- Harm Abstract, Noticed damages and severity evaluation
- Really useful actions
This was routinely generated simply after you uploaded the picture and the barcode.
How does it work?
Behind this Telegram bot, we even have an n8n workflow.
Harm Evaluation with Pc Imaginative and prescient utilizing n8n
Like within the earlier workflow, most nodes (in purple sticky notes) are used for orchestration and data assortment.
The workflow can also be triggered by messages obtained from the operator:
- (1) and (2) be sure that we ship the instruction message to the operator if the message doesn’t include a picture
- (3) is utilizing state variables to know if we count on to have an image of broken items or a barcode
The output is distributed to AI-powered blocks.
If we count on a barcode, the file is distributed to part (4); in any other case, it’s despatched to part (5).
For each, we’re utilizing OpenAI’s Analyze Picture nodes of n8n.
The downloaded picture is distributed to the picture evaluation node with a simple immediate.
Learn the barcode, simply output the worth, nothing else.Right here, I selected to make use of a generative AI mannequin as a result of we can’t assure that operators will all the time present clear bar code photographs.
For (5), the system immediate is barely extra superior to make sure the report is full.
You're an AI assistant specialised in warehouse operations
and damaged-goods reporting.
Analyze the picture supplied and output a clear, structured injury report.
Keep factual and describe solely what you may see.
Your output MUST observe this actual construction:
Harm Abstract:
- [1–2 sentence high-level description]
Noticed Harm:
- Packaging situation: [...]
- Pallet situation: [...]
- Product situation: [...]
- Stability: [...]
Severity: [Minor / Moderate / Severe]
Really useful Actions:
- [...]
- [...]
Pointers:
- Do NOT hallucinate data not seen within the picture.
- If one thing is unclear, write: "Not seen".
- Severity should be one among: Minor, Average, Extreme.This technique immediate was written in session with the standard staff, who shared their expectations for the report.
This report is saved in a state variable that shall be utilized by (6) and (7) to generate the e-mail.
Generate Report – (Picture by Samir Saci)The report consists of JavaScript code and an HTML template which are populated with the report knowledge and the barcode.
The ultimate result’s a concise report able to be despatched to our high quality staff.
If you wish to take a look at this workflow in your occasion, you may observe the detailed tutorial (+ template shared) on this video.
All these options will be straight carried out in your n8n occasion.
However what when you’ve got by no means used n8n?
Begin Studying Automation with n8n
For the inexperienced persons, I’ve ready an entire end-to-end tutorial through which I present you learn how to:
- Set your n8n occasion
- Arrange the credentials to connect with Google Sheets, Gmail and Telegram
- Carry out fundamental knowledge processing and create your first AI Agent Node
On the finish of this tutorial, it is possible for you to to run any of those workflows introduced above.
A good way to follow is to adapt them to your personal operations.
enhance this workflow?
I problem you to enhance this preliminary model utilizing the Textual content-To-Speech capabilities of generative AI fashions.
We are able to, for example, ask the operator to offer further context by way of audio and have an AI Agent node incorporate it into the report.
Conclusion
This isn’t my first challenge utilizing n8n to automate workflows and create AI-powered automations.
Nevertheless, these workflows had been all the time linked to complicated analytics merchandise performing optimisation (budget allocation, production planning) or forecasting.
These superior prescriptive analytics capabilities addressed the challenges confronted by massive corporations.
To assist much less mature SMEs, I needed to take a extra pragmatic method and deal with fixing “native points”.
That is what I attempted to exhibit right here.
I hope this was convincing sufficient. Don’t hesitate to strive the workflows your self utilizing my tutorials.
Within the subsequent article, we are going to discover utilizing an MCP server to reinforce these workflows.
Let’s join on LinkedIn and Twitter; I’m a Provide Chain Engineer utilizing knowledge analytics to enhance logistics operations and cut back prices.
For consulting or recommendation on analytics and sustainable provide chain transformation, be at liberty to contact me by way of Logigreen Consulting.
