What drove the decision to develop an AI-based use case?

Many companies have thousands of existing customers, but the main focus is usually on acquiring new ones. Yet existing customers are often easier to activate – they already know the company. Traditional methods like newsletters are no longer effective, as they feel impersonal and the target audience is saturated.

Personal outreach through sales would be ideal, but often fails due to lack of time. A middle ground is therefore personalized messaging via email and WhatsApp, powered by AI. This enables true personalization while maintaining high efficiency. That’s why we – together with our clients – chose exactly this path.

 

What were your expectations or requirements for the AI application?

AI often sounds like automation, but not like closeness. That’s why it was important to us: human and machine must always share a common understanding of the customer. Only then can AI shape communication in a meaningful way. Moreover, humans remain in control at all times – either by approving each message or via regular spot checks ("human-on-the-loop").

A second key point: measurability. The impact of the AI should not only be noticeable but also visible – through clear KPIs such as response rates and conversion rates.

 

Which partners and technologies did you rely on during development?

Compliance with the GDPR and the EU AI Act was a central requirement. Customer data was never allowed to be stored on non-European servers. We therefore rely on Microsoft as our technology partner, with AI model hosting in Sweden. Currently, we are working with Deutsche Telekom to ensure full data processing within Germany on Telekom servers – especially important for sensitive industries.

 

The use of AI solutions must also be accompanied by a change in competences. How did you handle the skill shift within the team?

Our initial team members were international experts in AI and CRM. New colleagues have grown into the topic from the start. Among our software customers, the key question early on is: what can AI do – and what not?

That’s why we emphasize transparency (also in line with the EU AI Act) and closely support our customers in the early stages. This helps build knowledge and trust in the technology.

 

In what timeframe did they implement the AI solution in your company/your client‘s company?

Development of our platform began at the end of 2023. In early 2024, we launched with the first customer projects.

Smaller companies are usually technically onboarded within 2–3 days. With coordination, it takes around 1–2 weeks until the first customer activation. For larger SMEs with complex IT setups and multiple stakeholders, the process can take up to 3 months.

From there, you can think about extensions and improvements – for example, integrating more data sources or expanding the use case to other parts of the organization

 

A conclusion to your AI solution:

Our customers work with us long-term – because we only take action where the business case and data situation are right.

The success is clearly measurable:
– AI activation generates up to 20x more order volume than traditional newsletters
– An average project yields 3x more profit than it costs
– Without complex system integration, the ROI is often achieved from the first month

Another positive outcome: Even customers who currently have no need feel well-addressed and appreciated – and often send back kind messages.

 

One final piece of advice/tip to other entrepreneurs looking to apply AI:

Not every “AI sticker” delivers real value. Scrutinize marketing promises – and do the math on the business case.

We can, for instance, very accurately assess whether implementation will pay off – based on margin, number of reactivatable customers, and expected conversion.

Important: "Don’t let the techies build it alone in the basement.” Sustainable, successful AI applications need more than just infrastructure, data, and models – they also need a focus on the application layer and user experience.

 

 

Further information:

http://www.mainition.de

https://www.youtube.com/watch?v=fuNYdq3rGAs&t=2s

 

Picture: Mainition