How-Tos & Practical Guides

Full autonomy is the goal for many agentic workflows — but full autonomy is also where most production deployments fail their first risk review. The practical path to deploying AI agents in real organizations runs through human-in-the-loop (HITL) patterns: workflows where the agent does the work, humans approve the decisions, and the system handles the handoff cleanly. LangGraph has strong native support for HITL patterns through its interrupt primitives. This guide walks through the core patterns — interrupt points, approval gates, and reversible actions — with working code you can adapt for your own agent workflows. ...

LangSmith Fleet formalizes two agent authorization models: Assistants (on-behalf-of user credentials) and Claws (fixed service-account credentials). Picking the wrong one creates either security gaps or broken functionality. This guide helps you choose and implement correctly. For background on why this distinction matters, see: LangChain Formalizes Two-Tier Agent Authorization in LangSmith Fleet Decision Framework: Which Model Do You Need? Answer these questions before you write a line of config: 1. Does the agent access data that belongs to the individual user interacting with it? ...

ByteDance open-sourced DeerFlow 2.0 on February 27, 2026 — a full SuperAgent harness rebuilt on LangGraph 1.0 that shipped with persistent memory, sandboxed execution, file system access, skills, and sub-agent support baked in. It hit GitHub Trending #1 within 24 hours and crossed 25,000+ stars in days. If you want to try a production-grade agent framework without building the plumbing yourself, DeerFlow 2.0 is one of the most complete starting points available right now. Here’s how to get it running locally. ...

One of the most consequential decisions in enterprise AI agent deployment is also one of the least discussed: should your agent act as the user, or as a service? LangChain’s Harrison Chase formalized this question in a March 23 post on “In the Loop,” LangChain’s developer newsletter, introducing the two-tier authorization model now available in LangSmith Fleet. The framework is called Assistants vs. Claws, and it directly addresses a security gap that enterprise teams have been quietly dealing with for months. ...

OpenClaw v2026.3.22 is a significant release — but if you installed it via npm, you may have hit two frustrating regressions immediately: a blank Dashboard UI and a WhatsApp plugin that simply doesn’t load. Both issues are fixable right now. Here’s what happened and exactly how to resolve each one. What Broke and Why Regression 1: Blank Dashboard UI Root cause: The npm tarball for v2026.3.22 is missing the dist/control-ui/ directory. This folder contains the compiled frontend assets for the Dashboard web interface. Without it, the Dashboard loads an empty shell. ...

OpenClaw v2026.3.22 ships ClawHub — a native plugin and skills marketplace built directly into the platform. No more hunting npm packages or manually dropping files into skill directories. ClawHub lets you browse, install, and manage extensions from a curated registry in one place. This guide walks you through everything: how ClawHub works, how to install your first skill, and a real-world example using Hitem3D — the first major third-party integration spotlighted on the marketplace — to generate 3D models from images. ...

A developer recently published an account of running 10 Claude Code agents simultaneously on their codebase — and the performance difference was not subtle. Analysis that previously took 10 minutes dropped to 3. If you’ve been running Claude Code agents serially, this guide covers exactly how to replicate that setup. Why Parallel Agents Work Claude Code’s Agent Teams architecture is built for parallelism. Each agent operates in its own context window with its own task scope, writing results to shared inboxes rather than competing for a single context. The bottleneck in serial workflows isn’t usually the model — it’s the sequential handoff pattern. ...

For most of OpenClaw’s history, backing up your configuration meant manually copying files, writing shell scripts, or hoping your dotfiles repo was up to date. With v2026.3.8, that changes: OpenClaw now ships with built-in backup CLI tools that make protecting your setup first-class behavior, not an afterthought. This guide covers the new openclaw backup commands, how to use config-only mode, how to verify your backups, and how to restore if things go wrong. ...

⚠️ Time-sensitive: Free API access for MiMo-V2-Pro expires approximately March 25, 2026. Xiaomi’s MiMo-V2-Pro is now live on OpenRouter, and for the next few days, you can run it for free. This is a frontier-class agentic model (1T parameters, sparse 42B active) that benchmarks close to Anthropic’s Opus 4.6 — and it was purpose-built for the kinds of autonomous, multi-step tasks that OpenClaw agents perform. Here’s how to hook it up in under 10 minutes. ...

Every AI agent pipeline eventually hits the same wall: documents. PDFs, Word files, scanned images, slide decks — agents need to read them all. Most solutions are either painfully slow, require an external API (and cloud costs), or demand a GPU just to process a 40-page report. LlamaIndex founder Jerry Liu announced LiteParse on X on March 19th, calling it “unglamorous but critical” infrastructure. He wasn’t wrong. LiteParse processes 500 pages in 2 seconds on CPU. No GPU. No API key. No cloud. ...