Gustav Norrelund

Speaking

I speak about how AI agents fail, how to deploy them anyway, and who is buying AI in Europe.

To book me, email gustav@northgrove.ai with the date, audience and format.

Topics

  1. How AI agents fail, and how to ship them anyway

    A live demo of attacks on an agent, the "lethal trifecta" of private data, untrusted input and outbound actions, and the design patterns that hold up.

  2. Who is buying AI in Europe

    Data from Follow the Stack on public AI contracts: who buys, who wins, and what it says about the market.

  3. Sovereign AI: rhetoric vs. reality

    From my thesis: what governments and companies say about sovereignty, and what they actually buy.

Formats

Press kit

Name: Gustav Norrelund. Pronounced “GUS-tav NOR-reh-lund”.

Short bio (50 words)

Gustav Norrelund builds and red-teams AI agents. He spent a year on Meta's Generative AI red team and has worked at Google and EY. He runs Northgrove, a studio that builds secure AI agents for sales and support teams, and writes Follow the Stack, a data publication on who buys AI in Europe.

Long bio (150 words)

Gustav Norrelund builds AI agents and finds the ways they break. He spent a year on Meta's Generative AI red team, testing how AI systems can be pushed into doing what they shouldn't, and has also worked at Google and EY. Today he runs Northgrove, a studio that builds AI agents for sales, support and customer success teams and red-teams them before they go live. He also writes Follow the Stack, a data publication tracking who buys and wins public AI contracts in Europe. Gustav is finishing an MSc in International Business and Politics at Copenhagen Business School, with a thesis on sovereign AI: the gap between what governments and companies say about technological sovereignty and what they actually buy from US hyperscalers. He speaks in English and Danish.

Downloads