Your own role
A close look at how you actually work: where AI makes your existing processes faster, where they are worth rethinking, and what is genuinely worth automating. For a single leader as much as for a small leadership team.
Training & workshops
For companies and universities
Workshops, talks and training, shaped around your organisation and your role.
The whole picture
At this size you can still see everything at once. We go through processes, tools and responsibilities together and decide where AI genuinely helps — and where it adds nothing.
One bottleneck, in depth
Instead of the full survey, a targeted look at the real pain point. Usually that means scaling AI across many business units: integration patterns, availability and choice of tooling, governance, and cost transparency.
Fundamentals that come up almost every time
- The lethal trifecta (Simon Willison): an agent turns dangerous the moment access to sensitive data, untrusted content and an outbound channel come together — and how to mitigate that combination in your own landscape, weighed against risk and value.
- Identity and access management for AI: how do AI agents authenticate? Do you accept impersonation or only delegation, and how do you spot risky combinations of access as they happen?
- Keeping an overview: which tools, MCP servers and agents exist in the first place, who runs them, who is allowed to use them, and how new ones are added under control.
- Integration patterns: when an API is enough, when you need an MCP server, and when it has to be an A2A agent. And when AI is the wrong answer.
- Governance that grows with you instead of slowing you down — guard rails that still work once more than a handful of teams are involved.
- Cost management: making consumption visible across the entire company by decoupling applications from model access.
- Shadow IT: why it appears particularly fast around AI, and how to meet it without shutting innovation down.
For universities and higher education
Lectures, seminars and guest talks
A deep look at the technology underneath, at lecture level. Also available to architecture and platform teams at engineering-led companies.
Identity and access management from the ground up
Authentication, authorization and federation basics — what the terms actually mean, the security principles behind them, and the common authorization models. Then OAuth 2 token flows in detail.
- Authorization Code with PKCE – the standard case wherever a user is present
- Client Credentials – service to service, with no user involved
- Device Authorization Flow – devices with neither a browser nor a keyboard
- Impersonation, delegation and token exchange (RFC 8693) – calling further services on a user's behalf
- More specialised flows: – RAR, PAR, JAR, JWE
The protocol landscape around agents
An overview of the protocols and what each one is actually for.
- REST APIs – agent to tools and data?
- MCP – agent to tools and data
- A2A – agent to agent, with the Linux Foundation since 2025
- A2UI, AG-UI – what an agent is allowed to render in the interface
- AP2 / x402 – mandate, approval and proof for agentic payments
- UCP – agentic commerce
Scaling problems from real environments
Why AI integrations so often fail on identity, governance and integration — and how to build agentic systems safely (OWASP Top 10 for Agentic Applications).
Ready to book, or would you like a tailored proposal?
Just drop me a line — I reply within 48 hours on working days. Format, scope and price all follow from what you want to achieve.