The most effective entry points for SMEs are tasks that occur regularly, are time-consuming, and have clear patterns:

• Communication: Drafting emails, wording quotes, creating social media copy.
• Document processing: Analysing contracts, extracting invoice data, summarising minutes.
• Customer service: Automatically answering frequently asked questions, categorising and prioritising enquiries.
• Knowledge management: Making internal knowledge searchable, preparing onboarding materials.
• Analysis: Evaluating customer feedback, identifying market trends, generating reports.

The key is: start small, demonstrate value quickly, then scale.

Costs vary widely depending on the approach:

• ChatGPT Team / Claude Pro: approx. USD 20–30 per user/month. Ready to use immediately, but limited customisation.
• API-based solutions: Billed by token usage. For typical SME applications, often between CHF 50 and CHF 500 per month, depending on volume.
• Custom solutions: One-off development costs plus ongoing operating costs. Makes sense when standard solutions are not sufficient.

Important: Do not compare the costs to “free”, but to the status quo. If an employee spends 10 hours per week on a task that AI can complete in 1 hour, the investment pays off quickly.

That depends on how you use AI:

• Free services (e.g. ChatGPT Free): Your inputs may be used to train future models. Not recommended for confidential business data.
• Paid API usage: With OpenAI, Anthropic, and Google, API data is not used for training by default. The data is deleted after a short time.
• On-premise / private cloud: The model runs on your own infrastructure. No data leaves your company.

At nuwai, we advise you on three security levels—from lean cloud integration to a fully isolated solution.

Yes—but it requires careful planning. The key points:

• Data processing: Clarify which data flows to which provider and where the servers are located. With US providers, the EU–US Data Privacy Framework must be taken into account.
• Transparency: Inform customers and employees when AI is being used. This is particularly relevant for automated decisions.
• Data processing agreement: Conclude a data processing agreement (AVV/DPA) with the AI provider.
• Data minimisation: Send only the data to the AI that is actually required.

At nuwai, we place great importance on ensuring that every solution is designed to be privacy-compliant from the outset.

On-premise (or on-prem) means that the AI software runs on your own servers or in your own cloud infrastructure—not with an external provider. Not a single byte of data leaves your company.

This is the most secure option and is particularly relevant for industries with strict compliance requirements: financial services, healthcare, public authorities, law firms. The downside: higher upfront investment in hardware and operations. In return, you have full control over data, availability, and costs.

As a rule, AI does not replace entire roles, but individual tasks within a role. The most accurate statement in our experience: “AI does not replace people—but people who use AI replace people who do not.”

In practice, especially in SMEs, we primarily see a shift: repetitive, time-consuming tasks are automated, while employees can focus on what requires human judgement, creativity, and relationship skills. The most sensible mindset is to view AI as a tool that enhances your work—not as competition.

Our recommended approach in three steps:

1. Understand: Identify the biggest time sinks and pain points in your operations. Where are things done manually even though they actually follow patterns?

2. Start: Choose a specific use case with clear benefits and manageable risk. Implement a quick pilot—ideally in weeks, not months.

3. Scale: Learn from the pilot, measure the benefits, and expand systematically.

The biggest mistake we see: companies waiting for the “perfect moment”. The technology is mature. The best time to start is now.