London Software Guild #2

About the London Software Guild
The London Software Guild gathers developers, architects, and teams across London who care deeply about the craft of building great software. No fluff, no vendor pitches — just people who take the work seriously, sharing what they've learned the hard way.

Agenda:

Talk: Dimming the Lights
Abstract: Lessons from the front lines of transforming our enterprise SDLC into a dark factory. We'll talk about quick wins, surprising sticking points, the unsolved problems of orchestration and validation, and controlled experiments and challenges that you can try with your team.
🎤 Speaker: Macey Baker, Community Engineer at @Tessl
Bio: As Tessl's Founding Community Engineer, Macey Baker is busy helping to build the future of AI-native development (and helping you build it too). Now an AI code generation obsessive, she got her start as an early employee at tech unicorn Intercom in San Francisco, before jumping the pond and roaming the London startup scene, working primarily in big data / ML houses. Some of her best friends are LLMs.

Talk: Secrets in Plain Sight
Abstract: Your AI assistant needs to understand your code, but your model provider doesn't need to know your credentials. We'll explore a novel guardrail architecture that lets models reason about sensitive data they never actually see — preserving full performance without the leak. We'll cover the design philosophy, the tradeoffs, and why context-aware approaches outperform blanket redaction.
🎤 Speaker: Daniel Trugman, Co-founder and CTO at Requesty
Requesty is building a novel AI Gateway that provides a complete control plane for any AI use case. The team at Requesty is building innovative solutions that solve the problems of tomorrow for AI trailblazers.

Talk: Agent Optimiser
Abstract: Agents are typically improved by testing them on a few tasks, observing what went wrong, updating the prompt, and repeating until the agent works on all tasks. However, this approach will only get you so far. It requires a human to be in the loop, or risk trusting your coding agent to evaluate. It's also unlikely to be optimal, there's a high chance you'll spend more money on tokens than you need to. This talk is about how he uses GEPA, the state-of-the-art prompt optimisation algorithm, to optimise conversational agents.
🎤 Speaker: Ben Clarke, Founder at @A2A Net
Ben is the founder of A2A Net, a platform to build and publish agents to Slack, Teams, Copilot, and Gemini Enterprise. Ben has an AI MSc from the University of Edinburgh and has been building websites, communities, and apps since he was 18. His latest focus has been on creating useful products from state-of-the-art AI research.