Agent-Memory

LangChain published a framework today for thinking about continual learning in AI agents — and it’s one of the clearest mental models for this problem that’s appeared in the wild. This guide takes that framework and turns it into a practical implementation playbook, with code examples for each layer and decision criteria for choosing between them. The three layers, briefly: agents can learn through context (runtime-injected instructions), storage (external memory), or weights (model fine-tuning). Each has different costs, speeds, and durability characteristics. ...

When developers talk about building AI agents that get smarter over time, they usually mean one of two very different things — and they rarely realize the ambiguity. LangChain’s Harrison Chase published a framework today that finally gives the field a shared vocabulary: continual learning for AI agents happens at three distinct layers, and conflating them leads to systems that are overbuilt for simple problems or structurally incapable of solving hard ones. ...

OpenClaw just shipped version 2026.3.28, and if you run agentic pipelines on this platform, you need to read the release notes carefully. This is one of the more architecturally significant updates in recent months — it introduces async human-in-the-loop (HITL) tool approvals, drops the Qwen portal auth integration entirely, and ships a handful of other meaningful improvements. Let’s unpack what changed and what it means for your deployments. Async Human-in-the-Loop: The Headline Feature The biggest change is the addition of requireApproval as an async hook in OpenClaw’s before_tool_call plugin system. In practical terms, this means plugins can now pause tool execution mid-flight and prompt the user for explicit approval before the tool actually runs. ...

Every developer who’s shipped an AI agent to production has run into the same wall: the agent remembers nothing across restarts. In-memory state is fine for demos. In production, where agents run for hours across multiple sessions, get killed by infrastructure failures, and need to pick up where they left off, in-memory state is a liability. Your agent’s entire conversational context, decision history, and accumulated knowledge evaporates the moment the process terminates. ...

Your LangGraph agent works perfectly in development. Then it hits production and you discover the problem every agent developer eventually hits: when the process restarts, your agent remembers nothing. In-memory state is fine for demos and local testing. For production agents — especially those handling multi-step workflows that can span hours, serve concurrent users, or need to resume after infrastructure failures — you need persistent state. This guide walks through adding Aerospike Database 8 as a durable memory store for your LangGraph agent. ...

OpenAI shipped ChatGPT Library — a persistent file storage system that survives across conversations — and most coverage has treated it as a quality-of-life feature. You can finally keep your documents without re-uploading them. Convenient! But there’s a more interesting way to read this announcement, and it’s the one that matters for anyone tracking how AI agents are evolving: this is memory infrastructure, and it’s the foundation that makes persistent agents possible at scale. ...

OpenAI has quietly shipped one of its most structurally important features in months: ChatGPT Library — persistent file storage that persists across conversations, available across ChatGPT’s web and app interfaces. On its surface, it looks like a convenience feature. Upload your documents, reference them later, organize them in one place. Useful, unremarkable. The analysis from Nicholas Rhodes in his Substack newsletter argues it’s actually something more significant: foundational long-term memory infrastructure for AI agents. ...

The memory problem in agentic AI is well understood: most agents are stateless. They start fresh every session, have no record of past decisions, and can’t explain why they did something three interactions ago. For demos, that’s fine. For production systems that need to audit, adapt, and coordinate over time, it’s a serious architectural gap. Context graphs are one of the most architecturally interesting answers to that problem — and FalkorDB’s recent technical breakdown is worth understanding even if you don’t use their specific product. ...