AI order processing in seconds
How Technotrading automated order processing across a 12,000-item catalog, with a single AI coworker plugged into Odoo.
Measurement period
3.5 months of production data
5,397 cases processed
0 sec
Per order
0
Cases processed
0
Items in catalog
0 min
Turnaround
We're incredibly happy with AgentsLab. After our first contact in September 2024 we went live with our AI coworker for order processing in November, and we're already hard at work on the next processes.
Jeroen van Waveren
CEO · Technotrading
01, Situation
A process that hung on a handful of people
Technotrading is an established Dutch technical distributor that supplies specialist products to business customers at home and abroad. Most orders come in by email, in wildly different formats: PDF, Excel, Word and plain text. On top of that, orders flow through their own webshop and an online marketplace. All told: more than a hundred orders a day.
The heart of the challenge sits in the catalog. Customers order in their own language, abbreviations, brand names, off-spec descriptions, industry-specific jargon. Matching a customer's request to the right product was a skill only the most experienced staff could consistently pull off.
100+
Orders per day
12,000
SKUs in catalog
4
Order formats (PDF/XLS/DOC/TXT)
The consequence
Overflowing inboxes, significant time pressure, and a process that depended on a few key people. Handling the tricky orders in particular takes a lot of experience.
02, Solution
AI order processing with one AI coworker, end-to-end
The AI coworker watches the order inbox, reads incoming orders in any format, and processes them fully automatically into Odoo.
01
Inbox monitoring
The AI coworker watches the order inbox and picks up every incoming message, PDF, Excel, Word or plain text.
02
Extraction
Customer details, product descriptions, quantities and references are pulled from the document in structured form.
03
Product matching
Customer-side descriptions, abbreviations and industry jargon get matched to the right SKU from a 12,000-item catalog.
04
Creation in Odoo
The sales order is created straight into Odoo, no middle layer, no parallel system.
05
Human-in-the-loop
When something's ambiguous or deviates, the order is flagged for a person. Every validation feeds the memory, per customer, per category.
Product matching
Previously the exclusive domain of senior staff. Now the entire team has access to the same match quality, regardless of experience level.
Human-in-the-loop
When in doubt, a person validates. The system learns from every interaction, the memory grows per customer, per product category.
Seamless integration
Straight into Odoo. No parallel system, no extra software. The inbox is always up to date.
03, Results
Before vs. now, live production data
Measured across 5,397 processed cases over 3.5 months.
KPI
Before
Now
Delta
Processing time per order
1 to 2 minutes
27 seconds
Turnaround (throughput)
1 to 2 hours
49 minutes
Human involvement
100%
87.5%
Time pressure on admin
High
Lower
Product matching (12,000 items)
Seniors only
Entire team
Note on human involvement. The 87.5% means staff still briefly validate most orders. That's a deliberate choice at this stage: the company is building trust with the system. As the memory grows per customer and product category, that number keeps dropping over time.
04, Timeline
From first conversation to production in two months
Sep 2024
First contact
Quick Scan: order inbox reviewed, processes mapped, scope locked in.
Oct 2024
Build & integration
AI coworker connected to Odoo. Product catalog of 12,000 SKUs indexed. First customers onboarded as a pilot.
Nov 2024
Live in production
Inbox monitored, first orders processed end-to-end. Human-in-the-loop dashboard live.
Q1 2025
Volume scaled
5,397 cases processed. Memory per customer and product category keeps growing.
2025
Quote processing live
Second AI coworker on the same platform: customer requests are converted automatically into quotes in Odoo. Same 12,000-SKU catalog, now on the sales side as well.
05, What this really means
Not replacing people, scaling the best.
Technotrading's story shows a pattern we see across mid-market companies: critical operational knowledge sits inside the heads of a few experienced people. When they're out or overloaded, the whole process stalls.
The AI coworker doesn't solve that by replacing people, it solves it by making the expertise of the best available to the whole team. Juniors now deliver the same quality as the most experienced colleagues. Inboxes no longer pile up. Customers get a faster confirmation.
And the system gets better the longer it runs, per customer, per product category, per order type, the memory grows stronger. That's what makes this not a one-off efficiency gain, but a structural improvement that deepens over time.
Similar process?
Find out what an AI coworker can do for your team.
We start with a Quick Scan. No commitments, just a concrete plan on the table within two weeks.