Cover Image for Engineering Memory for AI Coding Agents in PHP Applications.
Cover Image for Engineering Memory for AI Coding Agents in PHP Applications.
Avatar for PHP Talks Lagos
Presented by
PHP Talks Lagos

Engineering Memory for AI Coding Agents in PHP Applications.

Google Meet
Registration
Past Event
Welcome! To join the event, please register below.
About Event

​Join us for the next PHP Talks Lagos meetup for a practical session on engineering memory for AI coding agents in PHP applications. Your team learns a migration rule from a customer outage or rejects an approach after testing it. In a later session, a coding agent can repeat the same mistake if it has no record of that decision.

​Teams capture this knowledge in CLAUDE.md files, AGENTS.md files and decision logs. The engineering work includes deciding what to record and when an agent needs it. You also need to ensure the guidance still applies as the codebase changes, and that an agent's unverified assumptions don't become instructions for the rest of the team.

​Using a real Laravel codebase, Semeton Balogun will examine how to maintain useful project memory and give agents the relevant decisions at the right moment. The session will also include an experiment comparing the same agent on the same tasks, with and without memory, to measure its effect.

​What we'll cover

  • ​Recording architectural decisions and project constraints so agents can use them

  • ​Keeping memory accurate as the codebase and team decisions change

  • ​Giving an agent the relevant context at the point it needs it

  • ​Preventing an agent's guesses from becoming accepted team rules

  • ​Recognizing when to enforce a rule with a test

  • ​Comparing agent performance on the same tasks with and without memory

​What to expect

​Semeton will work through examples of the decisions a team needs an AI coding agent to respect in a real Laravel codebase. We will examine how to record those decisions, validate them, and make them available during a task.

​You will also see a measured comparison of agent behavior with and without memory. The session will give you a clearer basis for deciding what belongs in your project's memory files and what your test suite should enforce.

​Who should attend

  • ​PHP and Laravel developers using AI coding agents in existing projects

  • ​Engineers maintaining CLAUDE.md, AGENTS.md or project decision logs

  • ​Developers investigating repeated mistakes in agent-generated code

  • ​Engineering leads evaluating how agent memory affects their team's development workflow

​About the speaker

​Semeton Balogun is a backend and systems engineer who builds multi-tenant platforms for banks and enterprise SaaS. He is the creator of Agent Contract, a tool that enforces rules and task boundaries for AI coding agents. He also contributes to DIGIT HCM, an open-source platform supporting public-sector digital transformation in English-speaking West African communities.

​His experience spans backend engineering, system architecture, and technical leadership. He focuses on AI-assisted software development and agentic coding, emphasizing guardrails that help teams improve workflows while maintaining sound software engineering practices.

​About PHP Talks Lagos

​PHP Talks Lagos is a developer community focused on practical knowledge sharing around PHP and Laravel, including modern backend development. We host regular meetups where engineers learn together and share what's worked for them.

Avatar for PHP Talks Lagos
Presented by
PHP Talks Lagos