Services
AI & Software
Solutions
Training
Tools
Insights
References
Contact
Free initial consultation Ask a first question by email
About Deutsch
← AI & Software

Process mapping, as-is to to-be

Capture the process, design the target chain and decide per step whether a rule, automation, AI or a person fits. With effort, benefit and sequence.

A mid-sized company wants AI and starts with the tool. Three months later there is a chatbot nobody uses, because the real bottleneck sits somewhere else: in the handover from sales to order, in double entry, or in an approval nobody documents. Mapping the current process is not paperwork. It decides whether the investment lands.

How I work: with source · with approval · in your system.

Starting point

Most businesses know their processes from memory, not as a chain. In conversations, three versions of the same process appear: the official one, the lived one, and the one that survived in a spreadsheet. Those differences are the useful part.

What I do

I capture the process as it runs: conversations with the people involved, documents and screenshots, a look into the systems. The result is a chain from trigger to outcome, with volumes, times and handover points.

For each step I ask four questions: Can it be described as a rule? Is automation without a model enough? Does it need AI because text or unclear cases are involved? Or must a person decide? That assignment is the core of the work. A step that works as a rule does not need a model; a step with real judgement does not belong in automation.

Then I calculate per candidate: setup effort, running cost, time saved, risk. The sequence follows that ratio, not enthusiasm.

Result

You get a process map with as-is and to-be, a list of measures with effort and benefit per step, and a recommended sequence. The documents are structured so they can support a funding application.

Example

This preparatory work has been part of my project work since 2008: capturing requirements, modelling processes, documenting decisions. For e-invoicing and the DATEV export, exactly this chain was needed before formats and interfaces could be discussed. The concept phase can be fundable, subject to approval in each case.

Limit

A process map is not a project. It shows where it hurts and what a step would cost. Delivery then needs someone who builds and runs it. If the mapping shows that a step is cheaper as it is, the report says so.

Your effort is conversation time: whoever knows the process has to explain what they do and correct the record. Without that time, the map is built on guesses.

How it runs

  1. Initial call and scope: which process, which outcome
  2. Capture on site and in the systems, separated by role
  3. Design the to-be chain, calculate variants
  4. Present the result, hand over the decision paper
  5. On request: support the first measure through delivery

Common questions

How long does the mapping take?

Two to four weeks for a clearly scoped process. Longer for several sites or systems.

Do I need access to every system?

No. Conversations and documents often suffice. A look into the systems helps because that is where the lived exceptions become visible.

How is this different from an ERP discovery?

A discovery looks at the ERP implementation project. Process mapping looks at the flow independent of the system and asks which step should be automated at all.

Is it fundable?

The concept phase can be fundable. It depends on the programme, location and application. I provide the documents; the funding body decides.

The next step costs nothing.

30 to 60 minutes, no strings attached: where is the biggest friction, what is the most effective lever, and is a collaboration a fit?