Kai in everyday work – How do you find your Kai?

In the 15th part of our AI series, we explore how a craft business can search for the right digital assistant.

This article is part of the special topic " Artificial Intelligence in Craftsmanship – an assistant called KaI"

A total of six steps are necessary to use KaI – your digital assistant – effectively in your company:

1. Step | Observe | Analyze the current state
2. Step | Identify the need for action | Develop a strategy
3. Step | Define the challenge | Select suitable solutions
4. Step | Identify solution | Test solutions and processes in pilot projects
5. Step | Implementation | Implementation and use throughout the entire company
6. Step | Process Management | Scaling and Further Development

In this episode, we will look at the first three steps.

Operational inventory

First, it needs to be clear what you're looking for. You wouldn't hire a tile setter if you needed a roofer, would you? This requires an assessment of the current situation. 

• regarding the processes within the company;
• the hardware and software that is used – i.e., the existing infrastructure and the possibilities of connecting it with new tools, so that data consistency is ensured;
• the data/information necessary for a job, i.e., the corresponding skills and abilities for the tasks that need to be done.

Photo: © Mittelstand-Digital Center for Crafts

This inventory should end in a simple presentation – such as the graphic "Process steps of automated invoice verification".

You might also be interested in:

Operations management

Select, establish, and further develop AI

Operations management

Social media without stress

The Chambers of Skilled Crafts in Germany - HWK Münster

Between AI and securing skilled workers: Employees look to the future

Six W-questions

When creating a workflow, it's essential to answer the six "W" questions: Who does what , when , why , how , and with which tool? It's advisable to create this inventory as a team . Why? The process often involves a lot of discussion. This exchange clarifies what's working well, what isn't, and which tasks are particularly frustrating for employees . Experience shows that these are often routine tasks that, unfortunately, are frequently not performed with the necessary care, but KaI can provide excellent support in these areas.

So, if you have a job in mind for KaI, discussions with the team will likely reveal other opportunities, or you may find that the one you had in mind isn't needed after all, or has a lower priority. Furthermore, the challenges that need to be considered during implementation will be identified. 

Overview of the data situation

The inventory is complete when there is a thorough understanding of the task at hand, meaning an overview of all the steps required and a detailed examination of KaI's task itself. It is clear which challenges need to be addressed and what data is available, as KaI's knowledge is data-driven.

This data can originate from within the company , such as during invoice verification by AI (supplier quotes or framework agreements). It must be verified whether sufficient data is available to train AI or whether AI with the necessary expertise can be acquired. The quality and timeliness of the data are of paramount importance. The better the problem is understood, the more effectively AI can be selected and tested.

AI series and application examples All parts of the AI ​​series can be found in our special issue "Artificial Intelligence in Crafts – an assistant called KaI".

In addition, there are further articles about AI on handwerksblatt.de, for example the online articles "A mighty clever way to communicate" about the voicemail app with AI chatbot from the start-up "meiti", "AI – how craftsmen benefit in their everyday business"Or"ChatGPT: Save time and money with AI" from the theme special "Digital Crafts". 

Developing a strategy

Photo: © Mittelstand-Digital Center for Crafts

Regarding the selection process, I'd like to briefly review the various application areas (see graphic "Application Areas of AI"). AI can process images and audio, be creative, generate analyses and forecasts, and suggest optimizations. It can also control robots or systems, such as in autonomous driving. Which of these capabilities should be utilized? This is the first question to answer when making a selection. The answer is based on the requirements placed on AI. These requirements are often referred to as a requirements catalog.

Formulation of the target criteria

Now we move on to selecting possible implementations. Furthermore, success or target criteria must be defined to enable the evaluation of success or failure. The SMART approach is helpful in formulating these target criteria so that success or failure can be determined from them. SMART stands for specific , measurable , achievable , relevant , and time-bound . The precise meaning behind this will become clear from the control questions presented below:

• Specific: What exactly is to be achieved? What qualities are being sought? Where is the goal to be achieved? Who is involved?
• Measurable: How can goal achievement be measured? How exactly? When will I know that I have achieved the goal?
• Appropriate: Is the goal motivating? Is it accepted by those involved? Is it actively achievable through the project?
• Realistic: Is the desired goal achievable within the scope of the project? Is it feasible?
• Scheduleable: By when should the goal be achieved? Within what timeframe should the goal be achieved? Is the goal achievable within the project timeframe?

Here's another example of how to formulate goals for the challenge of receiving an incredibly high number of simple customer inquiries that are massively disrupting daily operations. "Goal: Implement an AI chatbot on the website to answer simple customer inquiries. Reduce response time by 70% and decrease disruptions by 50% within the first year. Simple customer inquiries relate to opening hours, …."

Criteria catalog for implementation

Especially when implementing AI – artificial intelligence-based assistance – within a company, criteria regarding technical feasibility must be formulated, such as which type of AI models are suitable and whether the necessary IT resources are available to implement AI. A cost-benefit analysis must also be considered. For this purpose, the costs for developing and implementing AI should be estimated by an independent expert and compared to the benefits. Crucially, the legal and ethical aspects of using artificial intelligence must be examined. Key terms are data protection , transparency , and fairness . Furthermore, it is essential to assess how the use of AI will affect existing work processes and employees . This includes, among other things, planning training and change management measures.

In this post, we looked at steps 1 to 3. In episode 16 of our AI series, we'll examine steps 4 to 6. This will cover the application process and hiring of KaI, as well as how we further develop KaI's skills.

The author was an employee of the Mittelstand-Digital Zentrum Handwerk (SME Digital Center for Crafts), a project funded by the Federal Ministry for Economic Affairs and Energy. From 2016 to June 2026, this center supported craft businesses and craft organizations in leveraging the opportunities offered by digital technologies, processes, and business models – free of charge, vendor-neutral, and nationwide. Since 2024, there has been a particular focus on artificial intelligence.

DHB now also available digitally! Simply click here and register for the digital German Crafts Journal (DHB)!