Saving Tokens in Agent-Based Dev Workflows

Ever feel like your AI coding agent burns through tokens way faster than it should? Come hear how we tackled that problem, without making developers slower in the process. We'll dig into why context size, not prompt length, is really what drives token consumption, and why switching sessions at the wrong moment can actually make things worse. You'll get concrete habits you can start using right away: when it's worth hitting /clear, why a short CLAUDE.md beats a long one, and why those unused MCP servers you forgot about are quietly costing you. We'll also talk about picking the right model for the job and writing sharper prompts, then look ahead at what's next: RAG-based indexing, cost-control plugins, organizing work with skills, and knowing when a smaller, cheaper model is actually the better choice. If you build with AI coding agents day to day, this one's for you. When: September 24, 17:00 Where: 1068 Budapest, Dózsa György út 84/B- Gloster Digital office