AI should deliver value – and strengthen people

What interests me about AI in business is how a technical possibility becomes a tangible improvement: fewer avoidable costs, more reliable workflows and more room for good customer work. For me, this also raises a second question: what do we do with the time we gain?

Two Handelsblatt articles bring these perspectives together for me. Claudia Obmann describes the economic use of AI at Ludwig Meister and Otto. Annika Keilen interviews professor and supervisory board member Yasmin Weiß about productivity, learning and human capabilities. My view is that an AI project needs a transparent economic assessment and a deliberate approach to how people work together.

What can actually be counted as a saving

According to Handelsblatt, Ludwig Meister reports annual savings of €36,000 from replacing previously purchased software with an internally developed brainstorming application. At Otto, around €10 million in potential savings relates to a planned expansion stage of returns analysis. The latter is explicitly a forecast. These company statements illustrate different types of benefit; they are not independently verified evidence of net savings that can generally be achieved. [1]

For my assessment, I would therefore consider three quantities separately: expenditure actually avoided, usable working capacity and improvements in quality. A discontinued licence can represent a specific reduction in spending. An hour of work saved, by contrast, only becomes a business benefit when it can be put to good use. Fewer incorrect quotations would be a separate quality improvement that would also need measuring.

Every calculation must also include the effort involved: development, integration with existing systems, model and licence costs, data maintenance, training, checking and rework. I would assess benefits and costs over the same period and avoid counting time savings both as lower personnel costs and as additional output.

Time gained needs a purpose

In the interview, Yasmin Weiß reports a self-calculated saving of three hours a day. She uses this time for productivity, private responsibilities and learning. At the same time, she warns against translating AI time savings exclusively into a more intensive workload. This is her personal experience and assessment, not a generally transferable time saving. [2]

For a company, I would turn this into a concrete design question: what should the capacity gained be used for? In sales, it could mean more thorough preparation for conversations, more time for difficult customer concerns or learning new ways of working. This use should be agreed with employees rather than silently squeezing ever more tasks into the same working day.

Alongside time and costs, I would therefore examine how the work changes: how much checking is needed? Do exceptions become manageable? Do new burdens arise? Do employees know how to challenge results and report errors?

A concrete starting point: from an incoming enquiry to a checked quotation draft

My illustrative proposal starts with a clearly defined sales task. Before choosing a tool, we jointly map the existing workflow: where is information missing, where is data entered twice, and where do questions or errors arise? Processing time and rework establish the starting point.

We then divide the task into steps that can be checked. AI could extract details from the customer enquiry, compare them with approved customer and product data, flag open issues and prepare a quotation draft. Technical suitability, prices and delivery commitments are approved by the people responsible. The draft does not automatically become a binding commitment.

In a limited pilot, we compare similar cases. We measure the entire workflow, including checks and corrections. Beforehand, we agree on quality requirements, responsibilities and criteria for adapting, ending or expanding the trial. IT and the relevant specialists examine data access, security and integration with existing systems.

For me, evaluation has two sides: has the process improved economically and in quality? And can employees work well with the new workflow? Only then would I propose expanding it. Ending a trial in good time can also be a sound decision.

My contribution: combining sales experience and AI knowledge

I combine international sales and leadership experience with current AI knowledge. Understanding different customer requirements, shaping organisational interfaces and aligning teams around shared goals have defined my professional background. They are also essential to making good use of AI.

This is where I contribute as a sales manager: clarifying business problems with users, making requirements understandable and bringing together sales, IT, controlling and other specialist departments. For me, facilitation and coordination mean making conflicting objectives visible, following up on decisions and responsibilities, and incorporating feedback from day-to-day work into the assessment.

My commercial perspective focuses on bringing customer needs, business objectives and reliable commitments together. Collaboration with technical and subject-matter specialists should turn a good idea into a verifiable workflow that works in everyday practice. Clear responsibilities and specialist approvals are part of that.

For a company, the value of this combination lies in aligning sales development with business objectives and the people who put it into practice. This is where I want to contribute my experience and AI knowledge: assessing opportunities, sharpening priorities and engaging teams in workable change.

A convincing AI project creates measurable value, keeps responsibility clear and gives the time gained a purpose. Good customer relationships continue to need people who listen, understand the context and act reliably.

Which process in your company could benefit from AI – and what would you use the time gained for?

Sources and context

[1] Claudia Obmann / Handelsblatt: “KI: So spart ein Unternehmen 198.000 Euro durch eine Anwendung”, 13 November 2026. Company examples and self-reported information; forecasts are distinguished from effects already described. The figure in the headline is not used here as a verified metric because its relationship to the costs described is not clearly traceable.
https://www.handelsblatt.com/karriere/ki-so-spart-ein-unternehmen-198.000-euro-durch-eine-anwendung/100264889.html

[2] Annika Keilen / Handelsblatt: “KI: Langfristig gibt es ernst zu nehmende Risiken für sehr viele von uns”, interview with Yasmin Weiß, 11 November 2026; first published on 6 November 2026. The interviewee's personal experiences, assessments and expectations about the future are not generally applicable evidence of effects.
https://www.handelsblatt.com/technik/ki/ki-langfristig-gibt-es-ernst-zu-nehmende-risiken-fuer-sehr-viele-von-uns/100250168.html

Both articles were read in full after signing in on 14 November 2026. The proposed steps and the description of my potential contribution are my own assessment.

How can AI support create a real advantage in everyday work? Yasmin Weiß explains why clear goals, suitable tools and a deliberate approach to using the time gained belong together.