Three-layer memory for AI agents, replacing single chat-log retrieval

Three-layer memory for AI agents, replacing single chat-log retrieval targets a $6.0B = 2.5M developer teams x $2,400 ACV (team-level memory infrastructure at $200/mo) total addressable market with medium saturation and a year-over-year growth rate of 35%+ driven by AI developer tools and agent adoption.

Key trends driving demand: Agent adoption -- more products embed autonomous agents which need persistent multi-session state, increasing demand for memory infrastructure; RAG maturity -- production patterns for retrieval-augmented generation make composable memory layers feasible and valuable; LLM context and tooling -- expanding context windows and cheaper embeddings enable hybrid short-term and long-term memory without exorbitant compute costs.