Memory built into a stateful agent runtime as a first-class subsystem.
Memory is built into a stateful agent runtime as a first-class subsystem, not a separate library you bolt on. The runtime manages when to store, what to keep, and when to inject context back in, often alongside skills, sessions, and user profiles. You get memory by adopting the whole agent, not by wiring a component into your own. The strongest member models memory as RAM/cache/disk that the agent itself manages inside its own reasoning loop, and can self-edit its core memory over long horizons.
The agent owns its memory. In every other family, memory is a component you call from outside the agent. Here, memory is part of the agent's cognition — it decides what to keep, what to evict, and how to rewrite its own behavior over time. That makes it the only family suited to long-horizon autonomous agents that should self-improve, and the trade-off is that you must adopt the whole runtime rather than slot a memory layer into your existing stack.
3 catalogued.
Letta
The agent itself manages its memory (RAM/cache/disk) inside its reasoning loop.
Best for: Long-running autonomous agents that manage their own memory · Memory-first coding agents (Letta Code, the current flagship)
View memory card →Nous Research
A self-improving, always-on agent from Nous Research with a built-in closed learning loop: agent-curated persistent memory, autonomous skill creation after complex tasks, skills that self-improve during use, FTS5 cross-session search, and a Honcho dialectic user model. Runs on a $5 VPS, a GPU cluster, or serverless (Modal/Daytona).
Best for: Operators wanting an always-on, self-improving personal agent whose memory (episodic + procedural + user model) is core to the loop
View memory card →MemTensor
Non-parametric self-evolving agent memory that applies runtime reinforcement learning on an episodic memory store. Instead of passive semantic matching (retrieve nearest neighbours and hope), MemRL uses environmental feedback signals to learn which past episode strategies are actually useful and promote them via a Two-Phase Retrieval mechanism — decoupling stable reasoning from the plastic memory. Agents improve from experience without weight updates or fine-tuning.
Best for: Research and agentic systems where agents repeatedly solve similar tasks and can provide environmental feedback (reward signals) to improve memory selection over time
View memory card →Top-down recommendations from the use-case playbook. Each names the one binding constraint that picks the tool, the primary pick (which may sit in another family when the case spans more than one), and runner-ups.
Binding constraint: Procedural learning — get better at a recurring task, update behavior from past wins.
Pick
LangMem — The only one with explicit procedural memory + prompt optimization from episodic successes (native if you're on LangGraph).
Runner-up
From Agentic Memory: Use-Case Playbook 2026 · last verified 2026-06-28
Last verified 2026-06-28 · updated by manual-stub