Hey there, My top two priorities for OpenClaw are, obviously, getting the job done, but also doing it while keeping token usage to a strict minimum so I don't blow my LLM budget. By using a local model and digging into the logs, I realized that the sheer number of skills and tools available to the agent was massively bloating my request context. After feeding those logs into Gemini, I spotted a bunch of unnecessary tools and skills that had nothing to do with my specific task. That got me thinking: it would be way more efficient to create task-specific agents by hand-picking only the tools they actually need. For some tasks, I managed to cut the token count by almost 10x! I’d love to know—what are your main strategies for optimizing token consumption?

Hi! I am super new to OpenClaw and I am excited to learn more about its capabilities and what I can build! First off, I think I already did two major no nos which was setting up OpenClaw on my main laptop and I also did not use a VPN or go to Tailscale like the course suggests. I watched a few Youtube channels and went from there until I found this group :) I haven't gotten to deep into using it yet. I have connected it to telegram and my CRM platform for my business but other than that, I have not given it set "jobs" yet. Should I uninstall it and start over or what are my options? Appreciate the advice!

OpenClaw 4.5 is one of the biggest updates yet with video and music gen directly inside agent conversations. This is huge for SMM, AI media gen sandboxes, and scaling content/data. The second big feature is the new Dreaming memory system, a background memory system with three phases that runs automatically while you work (or sleep). The 3 Phases: - Light: handles short-term recall - Deep: promotes important information into long-term memory - REM: surfaces 'lasting truths', things the agent determines you’ll always need to know with light, deep, and REM phases to build smarter long-term recall. Agents and agentic infra is leaning hard into memory recall and scaling. This will always be a massive part of smart systems so get yourself familiar with memory, context window, and token usage. We just published the updates on SS: https://open.substack.com/pub/moltbot/p/openclaw-just-dropped-video-music?utm_campaign=post-expanded-share&utm_medium=web Drop below what you're working on now that requires good long-term memory recall and what systems you have in place to crank this. P.S. I run an AI second brain using Obsidian + Notion system that's self learning and auto organises my notes into a directory. It's separate from my OpenClaw setups but useful for better integrating your note taking and thinking by leveraging AI.

Since many are using Openclaw already, just curious where you guys run it?

I enjoy following Felix on twitter - an agent ( https://x.com/FelixCraftAI ) who publishes diaries and technical details on how to build agents and earn a living. Results so far are pretty astounding! I hope my silicon chimps can grow up to do similar and help support me in years to come. My publishing business is incorporating these agent abilities so that when people write books now, we can help them launch agent versions of their characters at the same time. If only I knew what I was doing it would be great - I hope this community can help and share as I shall also be sure to do over time