Introduction to Harness Engineering
A framing session on why agent reliability lives in the harness around the model, covering runtime, tools, memory, security and observability.
A London Agentic AI event with Amazon at their London office focused on harness engineering: the runtime, tools, memory, security and observability layer around models that turns them into production-ready AI agents.

Hosts, sponsors and speaker organisations represented at this London Agentic AI stop.



This London Agentic AI stop digs into harness engineering, the practical discipline of building, deploying, testing and operating production-ready AI agents. If an agent is a model plus a harness, this evening is about the harness: the runtime systems, tool integrations, memory management, security protocols and observability frameworks that make agentic systems reliable.
Amazon hosts the community at their London office for a technical evening. The programme opens with an introduction to harness engineering from Anisha Malde, followed by a deep technical talk from AWS and a partner session, before a panel and audience Q&A on what it takes to run agents in production.
A concise briefing on the technical context behind the event: why the topic mattered to the London Agentic AI community and what sponsors, speakers and attendees came to discuss.
The Luma and Meetup pages confirm the title, Amazon LHR16 London venue, 15 July 2026 timing and approval-based Luma registration.
The published programme lists an intro from Anisha Malde, an AWS technical talk, a partner talk and a panel.
Two further speakers are still to be confirmed at the time of publishing.
What harness engineering is and why the layer around the model decides production reliability.
How runtime systems and tool integrations are designed for agentic workloads.
How memory management and security protocols shape dependable agent behaviour.
How observability frameworks help teams operate, debug and improve agents in production.
What AWS and partner practitioners have learned deploying production-grade agentic systems.
A framing session on why agent reliability lives in the harness around the model, covering runtime, tools, memory, security and observability.
An AWS technical session on building and operating agent harnesses for production workloads. Speaker to be confirmed.
A partner technical session on harness engineering in practice. Speaker to be confirmed.
A panel and audience Q&A on running agents in production, covering tooling, memory, security and observability decisions.
Original registration, archive and recording links for this London Agentic AI event.
The UK's original, high-signal Agentic AI community for AI engineers, agent builders, researchers, founders and technical leaders building production agents.
Subscribe on Luma →