iConference AG
AI that lands

AI on the phone — and it understands Bernese German.

We don't just try AI in theory, we put it to work on ourselves. This time: an AI phone assistant that answers incoming calls when no one picks up. We tested the solution from CleverAI, a Swiss provider. The result convinced us, and we are moving into a pilot.

«Hello, I've got a quick question.» «Of course — I'm here for you.»

It understands Bernese German

Not «standard German with a Swiss accent», but genuine Swiss German. In our case Bernese German, flawlessly. Anyone who has watched an automated phone system fail on dialect knows how rare that is. Alongside the common languages, the dialect comprehension is solid.

Clean conversation flow

The assistant answers with the company name, clarifies the request, asks about urgency and contact details — structured, calm, without forcing the caller into a rigid menu. At the end, a clean summary lands in our inbox by email. No «press 1».

Composed with objections and follow-up questions

This is where real maturity shows. Callers don't stick to the script: they interrupt, ask back, wander off. That's exactly where weak AI falls apart. Not here — the assistant picks up follow-up questions, stays calm and steers the conversation back, without sounding like a machine that lost the thread.

Why the threshold has been reached

That's the difference between «technically possible» and «landed in everyday life». A phone assistant doesn't have to be perfect — it has to reliably capture the request, so no call gets lost. That threshold is reached.

So we're now moving into a pilot — with real callers, not in a test lab. We'll report back once we know how it holds up in daily business.

AI on the phone is no longer a promise. It picks up — and even understands Bernese German.

We guide Swiss SMEs along this path — human, competent, holistic.

Read next

29 July 2026

Private AI accounts at work: ban first, then introduce properly

Customer data in a private AI account: who is liable under the revised Swiss data protection act, how thin the case law still is, and the two steps an SME takes now.

Read →
22 July 2026

The most insidious trap in AI answers: they sound utterly convincing but are factually wrong

We tested four AI models on a fictitious contract file. The most dangerous result was not the error, but how convincingly the weakest model went wrong.

Read →
22 July 2026

Method: four AI models on an M&A contract file

Method and findings for the model test: four AI models against a fictitious, anonymised contract file, measured deterministically against a solution key.

Read →