AI Agents Keynote –
When AI stops answering and starts acting.
Roman Weishäupl is a keynote speaker on AI agents and agentic AI. He shows what happens when artificial intelligence no longer just answers questions but plans and completes tasks on its own – and what that means for workflows, roles and accountability in your organisation.
AI agents are AI systems that are given a goal and plan the route themselves: they use tools, check intermediate results and deliver a finished outcome. In his AI agents keynote, Roman Weishäupl shows what this already means today – on request live through vibe coding, where software takes shape on stage without him typing a single line of code. He puts into perspective where agents really change work, where their limits are and which guardrails organisations need before they delegate tasks.
From chatbot to AI agent
Three stages, one difference: how much work does the human hand over? The keynote makes the leap from an answer to a completed task tangible.
Chatbot
You ask a question, the AI answers. What happens with the answer is up to you – and so is every further step.
Assistant
The AI sits inside the tool – in email, spreadsheets or code – and makes suggestions. You decide and execute.
Agent
You set a goal. The agent plans the steps, uses tools, checks its intermediate results and comes back with something finished.
That makes agentic AI less a question of the model than of delegation: which tasks do you hand over, what access does the agent get – and who checks the result?
What your audience
takes away
A clear picture of what AI agents can do today – and a plan for how your own team gets started.
Understand AI agents
What sets an agent apart from a chatbot and an assistant – in plain language, not tech jargon.
Where agents deliver today
Research, analysis, software prototypes, recurring workflows: what already works and what is still a demo.
Learn to delegate
An agent is only as good as its brief: describe goals clearly, check results, stay accountable.
Set guardrails
Which data, access and approvals an agent gets – and where a human makes the call.
From experiment to standard
How the experiments of individual Digital Explorers become something the whole organisation benefits from.
From the screen into the world
Agents in devices and robots: why AIoT and humanoid robotics are the next stage.
Agentic AI you can watch: live vibe coding
Roman doesn't just talk about agents – he works with them. In the AI Bootcamp he describes on stage, in natural language, what should be built. The AI builds it: live, visible to everyone, without a single typed line of code.
- Roman's speaker website was built this way: rebuilt in three days – without a designer, without a developer, without classical programming.
- A dispatcher now handles certain monotonous tasks in around 20 minutes instead of three hours – with a tool he built himself.
- A prototype is not yet a product: security, architecture and scaling still need experienced developers.
The stories behind it: From VCR to Vibe Coding · The Era of Product Guys
The AI agents keynote is one focus area of Roman Weishäupl's AI keynote. These pages and articles take you further.
AI Keynote Speaker
All focus areas, live video, stages and questions about the AI keynote.
View overview KEYNOTEAI & leadership
Shadow AI, Digital Explorers and safe learning spaces: what leadership needs to shape now.
View keynote KEYNOTEAI for the Mittelstand
Getting started without a mega-project: with your own people, short paths and a first use case.
View keynote ARTICLEThe Era of Product Guys
How vibe coding turns people with ideas into makers – and why a prototype is not yet a product.
Read the article ARTICLEFrom VCR to Vibe Coding
Why Roman started programming at 47 – and rebuilt his website in three days.
Read the article FORMATAI Bootcamp
Keynote, live vibe coding and hands-on workshop: the team builds for itself – from 8 to 1,000 people.
View formatAI agents and the keynote: frequently asked questions
What is an AI agent?
An AI agent is an AI system that doesn't just answer a question but pursues a goal: it breaks the task down into steps, uses tools such as search, files or software, checks intermediate results and delivers an outcome at the end. The human sets the goal and the guardrails.
What does agentic AI mean – and how is it different from a chatbot?
Agentic AI is the umbrella term for AI that acts on its own instead of only answering. A chatbot gives an answer and waits for the next question. An agent works through a task across many steps and decides for itself what to do next.
What is Roman Weishäupl's AI agents keynote about?
The keynote shows what AI agents can already do today, where they change work and where their limits are. On request, Roman Weishäupl makes this visible live – through vibe coding, software takes shape on stage without him typing a single line of code. From there he works out which tasks organisations can delegate and which guardrails they need to do so.
Does the audience need technical knowledge?
No. Roman explains AI agents in plain language and with examples from practice. The keynote works for boards as well as for broad company audiences; for IT and digital teams it can go deeper technically.
Will AI agents replace employees?
Agents take over tasks, not accountability. In the keynote, Roman shows how roles shift: away from executing individual steps, towards describing goals and checking results. And he shows the limit: a prototype built by AI is not yet a finished product – security, quality and scaling still need experienced professionals.
How long is the keynote and can it be combined?
The AI agents keynote runs 45 to 90 minutes, with Q&A on request, in English or German. It can be booked as a focus area of the AI keynote or combined with the AI Bootcamp, where the team builds with AI itself. Ideally, get in touch 6–8 weeks before the event.
Book the AI agents keynote
As a standalone keynote, as a focus area of the AI keynote or combined with the AI Bootcamp – tailored to your event.
Or call +49 40 524 766 60