It has been a viceral experience how quickly the AI space has shifted. In my notes on AI learning 2024 topics were focussed on machine learning, forecasting, data pipelines, mlops, tensorflow and computer vision. The introduction of LLM has shifted AI learning in 2025 into the realm of prompt engineering, RAG, context engineering and just a little bit of tool calling and Model Context Protocol, where in 2026 we have to figure out how to orchestrate multi-agent setups, effectively govern work breakdowns to be handled by sub agents and optimise our model and token use by reducing context size and routing appropriate work to the least expensive models necessary to complete the sub tasks.
topics
- Agentic Engineering
- Multi Agent
- Role specialisation
- Swarm optimisation
- Token optimisation