Success Story

How Tyrolit automates customer orders with Tycom Flow

The success story shows how Tyrolit Tycom Flow - Customer Service deploys in a production environment, what results are measured and how work in customer service has changed.

View results

Results at a glance

level of automation

55%

the orders processed without manual modification up to SAP

recognition quality

96,9%

correctly recognised and mapped fields

Scalability potential

~100.000

orders to be created per year at full roll-out.
Currently approx. 130 orders / day.

Time-to-Value

6 weeks

from project launch to go-live

The project success

Results from Tyrolit production operations

The focus is on around 100,000 customer orders to be created per year, which in the medium term will be fully processed with Tycom Flow need to be automated and are received in various formats. With Tycom Flow, the relevant order data is automatically recognised, validated against existing SAP data and made ready for further order processing.

Since the introduction, staff no longer have to manually enter every order in full. They focus on discrepancies, incomplete information and technically demanding exception cases.
Currently, around 55 % of incoming orders reach the SAP system without any manual intervention. An order is only considered automated if not a single field was manually adjusted – even a single modified detail, such as a delivery date deliberately optimised by Tyrolit, counts as manual processing.

The project shows how a clearly defined AI use case can be integrated directly into an existing SAP landscape and evaluated using operational key figures. The value is generated not through an isolated AI application, but through the interaction of document comprehension, SAP master data, business rules and expert control.

Find out more about the Tycom Flow – Customer Service Module

Project profile

The Tyrolit project at a glance

The company

Tyrolit Group
Solution

Tycom Flow - Customer Service
SAP landscape

SAP S/4HANA 2023
input channels

Email and manual upload
(email text, PDFs, Excel/spreadsheets, images, etc.)
project duration

6 weeks until go-live
Go-Live

Mid-August 2026
Project scope

Live: DK, UK, US, HU, BR, CA, PL
In planning: DE, AT, FI, SE, NO, EE, IT, FR, BE, CZ, CN, ES, PT, UAE, IN, JP, KR
(As of Sep 2026)

Initial situation

The initial situation at Tyrolit

In Tyrolit's customer service, numerous details from customer orders need to be entered completely and correctly into SAP. Different formats, customer-specific details and recurring data entry made processing time-consuming and hindered scaling as volume grew.

Before the introduction of Tycom Flow, the relevant data was predominantly transferred manually. Consignment withdrawals and replenishments, as well as rework orders in particular, caused additional checks, queries and corrections.

Manual entry

Order data mostly had to be extracted from the incoming documents and entered into SAP.

Many document variants

Orders arrived via email (both internally and externally) in various formats – plain email text, PDFs, spreadsheets and images – with different structures, languages and item tables.

Complex examinations

Items, quantities, prices, delivery dates and customer-specific master data had to be checked prior to creation.

Project objectives

Objectives of the Tyrolit project

Tyrolit wanted to speed up order processing and develop the process further in a targeted manner. Recurring and clearly verifiable tasks were to be automated. In the case of unclear or business-critical information, the decision should remain with the responsible employee.

It was also essential that the solution integrated directly into the existing SAP process. The business unit should neither have to work with an additional standalone system nor transfer data between different applications.

Reduce manual data entry
Provide orders more quickly in SAP
Avoid transmission errors
Manage peak workloads better
Handle exceptions transparently
Establish a foundation for further AI use cases

Trial at Tyrolit

Implementation of the process at Tyrolit

In the project, the path from receipt of a customer order to processing in SAP was implemented using real Tyrolit documents and known exception cases. The following steps describe the project-specific procedure. The general functionality explains the Tycom Flow product page.

Understand document

The AI interprets orders from various channels regardless of format, structure and language, and recognises the relevant order information.

Validate data

Customers, materials, quantities, prices and further details are reconciled with the existing SAP master and transaction data.

Check for exceptions

Ambiguous or differing details are flagged. Employees specifically process only those items where a technical check is necessary.

Create order

For every decision, the underlying evidence is disclosed and assigned a confidence value. From a configurable threshold onwards, successfully classified orders are automatically created in the SAP system.

Implementation

The implementation in Tyrolit production operations

At the beginning, Tyrolit and Tycom analysed typical document variants, relevant data fields, known deviations and the required SAP information. In doing so, they consciously considered not just ideal standard cases.

Multi-page orders, different table structures, customer-specific article numbers and incomplete information were incorporated into testing and optimisation.

1 – Process analysis

Record document paths, manual steps, validation logic and exception cases.
2 – Data and rules

Define relevant fields, SAP master data and business validations.
3 – Integration and Testing

Connect Tycom Flow to SAP S/4HANA and test it with real customer orders.
4 – Technical validation

Review results together with customer service and purposefully refine the rules.
5 – Go-Live

Go-live for the first sales organisations after a six-week test phase – despite the holiday period and with results well above expectations.

Measured impact

Results after go-live

Since the go-live in mid-August 2026, Tycom Flow has already achieved around Tyrolit 1,000 customer orders processed – currently, around 130 orders are added per day, with a weekly increasing trend.

Around 55 % of these are processed fully automatically in SAP. For the remaining roughly 45 %, at least one manual adjustment is required; even a single changed field entry counts as manually processed.

The recognition quality currently stands at around 96 % at head level and 96.9 % at position level. The average processing time per document fell from around four minutes to about 20 seconds. Since then, customer service has focused on verification tasks and targeted individual adjustments instead of data entry.

key figure	Before	With Tycom Flow
Processing time per order	approx. 4 mins.	approx. 20 secs.
Automated orders	0 %	55 %
Orders with manual adjustment	100 %	45 %
measuring period	Mid-Aug 2026	ongoing (As of Sept 2026)

Economic benefit

Economic saving potential

Beyond the time saved, the benefits can also be quantified economically. The following deliberately conservative projection applies the measured processing time (around four minutes down to approximately 20 seconds per order) to the target volume once the full roll-out is complete.

Calculation assumptions: 100,000 orders to be created per year with a complete roll-out, average time saving of 3 minutes and 40 seconds per order, fully absorbed hourly rate of 50 euros and realization rate of 85 per cent. The projection describes a calculated potential and not a cost saving that has already been realised.

PROJECTION · FULL ROLL-OUT

Around 100,000 orders per year

basis of calculation for the planned full expansion

Time saved per order approx. 3 mins 40 secs.

Released capacity (gross) ~6,100 hrs/year

Realistically usable (85 %) ~5,200 hrs/year

Calculated savings potential

~€360,000/year

at €50/h total cost rate · mathematically equivalent to around three full-time staff

Change in the working day

Customer services handles exceptions instead of typing in data

Automation does not just change metrics. Recurring data entry tasks are handled by Tycom Flow. Employees primarily review cases where information is missing, unclear, or deviates from the stored SAP data.

Today the team is focusing on:

Technical clarification of incomplete or contradictory information
Customer-related decisions and queries
Targeted maintenance of relevant master and process data
management of business-critical exceptions
Use of direct SAP navigation and the very good field recognition – even with custom fields such as US forwarding agent numbers

Customer Service Voice

„Before Tycom Flow, our teams spent hours manually processing orders. Today, OCR and AI do the heavy lifting, improving accuracy, saving time, and allowing our people to focus on customers rather than administration.“

Global Head of Customer Care, Tyrolit Group

Exceptions in the establishment

Technical review of exception cases

Tycom Flow automates clearly defined process steps and uses SAP data for validation. Uncertain or business-critical cases remain transparent and are reviewed by the business department.

This makes order entry a measurable AI use case within the existing SAP process. Recognition quality, exception rate and business impact can be evaluated separately.

Human-in-the-loop
Automation with targeted control of deviations.
Measurable quality
Recognition and automation rates are continually evaluated.
SAP integrates
No isolated AI application alongside the operational process.
Expandable
Applicable to further document variants and processes.

Success factors and learnings

Learnings from the Tyrolit project

For a high degree of automation, document recognition alone was not enough. In the project, the interaction between real order variants, master data quality, business rules, SAP integration and expert control was crucial.

involve the department early

Real document variants and known exceptional cases were incorporated into the implementation from the very beginning.

Anticipating SAP data

Document recognition and master data validation were designed as a joint process.

correctly classifying the degree of automation

A correctly recognised order for which Tyrolit deliberately sets an optimised delivery date counts as manual in the statistics. The automation rate shown therefore deliberately reflects only completely unchanged orders.

The outlook

A basis for further document processes

The basis created in the project can be used in future for further countries, companies and document types. A phased roll-out to the planned countries (including DE, AT, FI, SE, NO, EE, IT, FR, BE, CZ, CN, ES, PT, UAE, IN, JP, KR) is scheduled with the aim of around 400 orders per day or roughly 100,000 orders per year.

With this, Tyrolit creates the prerequisite, to intelligently automate further document-based processes and the to purposefully enhance the existing SAP landscape with AI-powered functions.

More about Tycom Flow Accounts Payable

Daniel Testor

Cloud Architect

„Automation begins where systems understand content - not just read out fields.“

Test Tycom Flow with your own orders

Bring a real customer order. We will show you live how Tycom Flow recognises the data, matches it with SAP and prepares it for order creation.

Choose a suitable date.

Frequently asked questions about the Tyrolit success story

Which customer orders were included in the evaluation?

Around 1,000 productive customer orders from DK, UK, US, HU, BR, CA and PL were evaluated in the period since the go-live in mid-August 2026.

What level of automation has been achieved at Tyrolit?

During the period under review, 55 % of the customer orders included were processed automatically. An order is only considered automated if no field was manually adjusted prior to handover to SAP. Even a single change means the order is counted as manually processed.

How was the change in processing time measured?

The average processing time per order before and after the go-live was compared based on internal measurements, discussions with employees and coordination with an external partner.

Which cases does customer service still examine manually?

Manually checked are differing or incomplete details, incorrect data on the order as well as delivery dates adjusted by Tyrolit. Tycom Flow makes these cases visible and transfers them in a targeted manner to the relevant department.

Which areas were involved in the Tyrolit project?

The project scope included the customer service of the live sales organisations and the SAP S/4HANA environment.

How long did the implementation at Tyrolit take?

Six weeks elapsed from the start of the project to go-live. The testing phase began in early July 2026, and the production launch took place in mid-August 2026. The scope included the first productive sales organisations (including DK, UK, US).

The evaluation takes into account productive operations since mid-August 2026 as of September 2026.