Speaking · agentGreg
Talks and workshops
I take one measured result and make it usable by the people who have to decide something about it, whether they write the code or sign for it.
What I talk about
Agents that survive contact with production
What actually separates a convincing demo from a system that answers real customers at three in the morning: where to put a human, how escalation earns its keep, and which failures you have to design for rather than prompt away.
for product and engineering leadership
Where model reliability ends
How to read what a model knows about its own knowledge, what evaluation and detection can prove about a single output, and how to write that limit into a policy instead of discovering it in an incident review.
for teams shipping LLM features, and for anyone drafting an AI policy
Local-language and sovereign models
What changes when the model speaks your language natively rather than in translation, measured on Polish models rather than argued from principle, and what it costs to run one on your own hardware.
for Polish and Nordic teams, public sector, and university programmes
Appearances
- Can you prove that AI wrote this text? Aug 2026 · AI Ninjas
-
AI agents on RCS and SMS, with SerwerSMS.pl Jun 2026 · webinar
-
Eskadra Bielika, Mission Two: build a RAG system on a Polish model Jun 2026 · Gliwice
Looking for a speaker or a hands-on workshop for your team, meetup or conference? Write to me on LinkedIn and tell me who will be in the room. I speak in Polish and in English.