A newsletter covering several system design topics. The main piece explains three AI agent communication protocols: MCP (agent-to-tool), A2A (agent-to-agent), and ACP (REST-based agent-to-agent, now merged into A2A), noting MCP and A2A are complementary in production. Additional sections cover evaluation strategies for LLMs, RAG pipelines, coding agents, and multi-agent systems; a roundup of eight notable open-source frontier models; a high-level overview of distributed tracing with OpenTelemetry; and an explanation of Redis persistence mechanisms (AOF, RDB, and mixed approach).
Table of contents
Your AI agent has a blind spot (Sponsored)MCP vs A2A vs ACP: How AI Agents Actually Talk to Each OtherIntroducing Attio: the agentic CRM. (Sponsored)LLM vs RAG vs Agent evals8 Frontier Open Models I’m Most Excited AboutHow Distributed Tracing Works at the High Level?The Life of a Redis Query1.5K Impressions