💡 Use Cases
Real-world examples and success stories
Dev Replaces $239.99/Year Plaud Subscription With Local Transcription Skill
A developer replicated the Plaud device's paid transcription service using a locally-run app, skill, and cron job — skipping the $239.99/year subscription. Transcripts land on the local machine where they can be queried across past meetings.
War Radar: An Automated Pipeline That Uses AI to Classify Conflict and Evacuation Reports
War Radar publishes a daily world map of armed-conflict risk: a formula sets the risk level, AI models handle only narrow classification, and the project publishes how often each level was followed by a crisis.
OpenClaw Yolo Mode: Autonomous Agent Loop on a €20/Month Budget
A user wired OpenClaw into a Yolo Mode loop where a main agent assigns work to a Project Manager sub-agent, checks subscription usage hourly, and pauses everything below 20% until the limit resets.
OpenClaw Agent Broke at 65% Context: 683k Tokens, Zero Cache Reads on Ollama/GLM
A 23-hour OpenClaw Discord agent run on GLM 5.3 Flash via Ollama Cloud broke at 683k tokens — 65% of the configured 1M window — with no overflow error, no timeout, and cacheRead=0 on every affected turn.
Novice Builds Local 'Second Brain' on Intel Arc Pro B60: Qwen 3.8 27B at 38 tok/s, 128K Context
A non-developer runs OpenClaw as a personal agent on an old Dell, then builds a separate AI server around an Intel Arc Pro B60 serving Qwen 3.8 27B at ~38 tok/s with a 128K context window.
QWEN3.6-27B-MLX-8bit Matches 122B for Local Workflows at 29.5GB
A user running multi-hour local workflows says QWEN3.6-27B-MLX-8bit (29.5GB) matched a 79GB model on orchestration, research, and writing — just slower — freeing 50GB of RAM.
OpenClaw Agent's Plain MEMORY.md Setup Beats Memory Startup's Runtime in Temporal Test
A memory startup's runtime returned the wrong version in a temporal-decision test, tying all three at 1.000 relevance. The user's OpenClaw agent, with memory as markdown in a git repo, returned the current decision dated.
Hermes vs OpenClaw: Benchmark on Real Work Ends with a Memory Lesson
A user benchmarked Hermes and OpenClaw on real tasks, finding performance similar but OpenClaw's explicit memory won them over after Hermes over-generalized a passing compliment.
Triaging Community PRs with Multiplayer Agents in OpenClaw 2.0
Patrick Erichsen (OpenClaw Foundation) shows how to use multiplayer agents in OpenClaw 2.0 to triage and review community PRs. The automation checks Discord every 15 minutes, assigns a maintainer, and starts a shared agent session.
OpenClaw Agent Discovers 10 Circle Packing Records, Gets Cited on Math Site
A Reddit user's OpenClaw agent, MoltFire, discovered 10 new circle packing records overnight, earning a citation on Packomania next to Google DeepMind.
Thread per Project, File for the Long Stuff: How I Organize My OpenClaw Agent Home
An OpenClaw user rebuilds their agent home three times, landing on Discord with a thread-per-project structure and a file-baton workaround for long messages.
OpenClaw and Local Models: A User's Struggle with GPT Blocks and Seeking a Local Setup
A user details 5-day GPT blocks, fails with LM Studio, and seeks advice for running local models on RTX 5070 Ti + 4060.