How is a keynote booked, and how fast do you answer?

Through Athenas, who manage my calendar and fees: kontakt@athenas.de, or the form on the booking page. They answer inside the working day and scope the brief with me directly once the basics are settled.

Will you pitch your company from the stage?

No. There is no Aiomics slide, no product demo and no sales follow-up. I disclose the company as an interest at the start — it is the reason the examples are current — and that is where it ends. This matters for pharma compliance teams, so it is written down rather than left to be asked.

What languages, and are they equivalent?

German, English and French. All formats run in all three. German and English are equally strong; French works well for keynotes and prepared sessions, and for a workshop with heavy open discussion I would suggest German or English if the room can carry it.

What do you need from us technically?

A lapel or headset microphone, HDMI or USB-C to the stage system, 16:9 at 1920×1080, and my own laptop on stage. Confidence monitor appreciated, not required. I bring a clicker and adapters. For broadcast, a wired connection and a rehearsal slot the day before.

For insurance, banking, energy, industry and the public sector

Move fast and break things is not available to you.

Every reference case in the AI conversation comes from an industry that can ship a bad answer and apologise later. Yours cannot. I work in the one place with a stricter standard than yours, and the method travels.

Enquire about a date →

Dr. Sven Jungmann — Leadership conference

The pattern across regulated sectors

Governance built as a gate, not a lane

When the only answer legal can give is no, the work moves into private accounts and unapproved tools. The exposure goes up, not down, and now it is invisible.

Pilots that cannot cross the line into production

Everything works in the sandbox. What kills it is audit trails, retention rules, explainability to a supervisor, and who signs. Those are design constraints, not paperwork.

The wrong people being trained

Analysts get tool training. The people whose judgement is now the bottleneck — team leads, reviewers, underwriters, engineers who sign off — get a webinar.

Sven really brings fresh solutions to old problems. You can see he is pushing AI to its edges for very pragmatic problem-solving.

Fund manager · Private equity

Talks for this audience

When being wrong is expensive.

AI in industries that cannot move fast and break things.

EN · DE · FR

Built for regulated sectors: how to design an AI capability that a supervisor, an auditor or a court can inspect afterwards — without governance so heavy that the work goes underground. Uses GDPR and the EU AI Act as concrete cases rather than a compliance lecture.

What the room leaves with

Governance as a lane instead of a gate, with the design rules that make it one
The four production constraints that kill sandbox projects, known up front
How to tell a real regulatory blocker from an inherited assumption

Two senior roles, one AI, and me.

What changed when I took over two experts’ work with AI, and what it means for how you lead.

EN · DE · FR

In my own company I let two people go whose work I could not have done myself before, a lawyer in the COO role and a UI designer with healthcare experience. I did not refill the roles. I took over their jobs, with AI: sixteen hours a week at first, a normal week now, and more than €300,000 a year in salaries. The talk tells that story without pride, including the mistakes of the first months, and draws out what holds for any leader: which work AI lets you take back, which stays with the team, and why delegating is getting more expensive while doing it yourself is getting cheaper.

What the room leaves with

One criterion for deciding which tasks to take over with AI and which to leave in the team
Why productivity drops in the first weeks and how to get through that phase
What changes for leaders when two layers between them and the work disappear

More about this talk →

AI agents: when software starts to act.

What agents do today, where they fail, and how to set their limits.

EN · DE · FR

2026 is the year AI stops answering and starts doing: research, documents, coordination, whole process steps. My company builds systems in which several agents check each other’s results, for hospitals where an unnoticed error is expensive. The talk shows from that work what agents reliably do today, where they fail, and which controls an organisation puts in place before it lets agents near customer data, contracts or payments. With a straight answer to the question every room asks: how much does the AI still make up, and what do you do about it?

What the room leaves with

A map: which tasks agents take over today, which in twelve months, which not for a long time
The three controls without which no agent belongs near real data
How to settle responsibility when the software decides

More about this talk →

Written on this

Founder Mode Has a New Floor

The advice founders are given — hire great people, get out of the way — is the advice that nearly broke Airbnb. What changes when AI removes the excuse for not staying close to the work.

Read the essay →

Leadership offsites, risk committees, annual conferences.

If your compliance function has ever been called the department of no, this is the talk to put in front of both sides of that argument at once.

Enquire about a date → Or write directly: kontakt@athenas.de