I spent decades helping enterprises manage infrastructure, cloud, data protection and cyber resilience. AI is changing the operating model. My focus now is Mastering AI — learning how humans and businesses can use AI to create measurable outcomes. Here I explore: 🤖 AI agents & AI workforce ⚡ AI automation & productivity 🧠 Practical AI adoption 🔐 Secure enterprise AI & Zero Trust 🛡️ Data resilience, Backup & DR The goal isn't to use more AI. The goal is to accomplish more with AI — safely, securely and resiliently. Follow along as I learn, test and share what actually works.

🆓 New FREE Coding Model on OpenRouter! "recursive/flint-v1.0:free" ✅ $0 input/output ✅ 262K context ✅ 40B MoE (~2.5B active) Promising for Agent37 Hermes coding agents. Benchmark tool calling, code fixes and multi-step reliability before production. openrouter.ai/recursive/flin…

Which future wins? A) One super-powerful AI assistant B) A team of specialised AI agents @NousResearch Hermes → open agent orchestration @grok Bot → autonomous AI workers @ChatGPT Dots → persistent knowledge agents @Meta Muse → personal autonomous agent The winners won't necessarily be the people with the “best AI model.” They'll be the people who know how to orchestrate AI effectively. Prompting is becoming the starting point—not the destination. For AI Master Training, I think agent management will become a core workplace skill. Agree or disagree? Would you rather have one genius AI or five specialised AI agents?

🚨 “OpenRouter for AI media” may have just arrived. Image & video AI is getting messy: Runware. fal. WaveSpeed. Higgsfield. Different APIs. Different pricing. Different limits. Comfy Router changes that. One API can access frontier image, video, 3D & audio models — while letting you choose which provider actually runs the job. Why this matters: → Switch provider when another is cheaper → Avoid building multiple SDK integrations → Queue jobs instead of repeatedly handling 429 limits → One key + one job interface → No subscription required — usage is credit-based Today, you still choose the provider manually. But Comfy is already planning: ⚡ "fast_start" 🏎️ "fast_finish" 💰 "lowest_cost" 🛡️ "reliable" 🔄 ordered fallback That’s where it gets interesting. For AI agents like Agent37 Hermes, the future may not be choosing the “best model.” It may be choosing the best route to the model — dynamically based on cost, speed and availability. Would you use a media router instead of connecting directly to Runware/fal/etc.?

“We have backup.” Good. But that does not answer the most important #ransomware question: **Can you #recover cleanly, quickly and predictably?** At Arrosoft, we look beyond #backup: 🛡 Resist 🔎 Reveal ♻ Recover Because ransomware #resilience is not about whether you can restore a file. It is about whether you can #restore the **business**. If your primary environment was #compromised today, how #confident are you that your recovery would actually work? 🔗 linkedin.com/in/KokLeong #RansomwareRecovery #CyberResilience #DataProtection #Arrosoft

AI is changing the cloud faster than most security strategies can keep up. Cloud. AI. Zero Trust. The question is no longer whether security must evolve — it’s how fast we can learn, adapt and build together. That’s why community matters. Join the Cloud Security Alliance Malaysia Chapter and connect with practitioners, leaders and builders shaping the future of secure cloud and AI in Malaysia. Watch the video. Join the conversation. Be part of the community. 🌐 CloudSecurityAlliance.org #CloudSecurity #CyberSecurity #AIsecurity #ZeroTrust #CloudComputing #CSAMalaysia #MalaysiaTech

🚀 GLM-5.3 is now live on @TokenRouter_US own GPU cluster. For a limited time, users can access a FREE GLM-5.3 model to explore its capabilities and test workloads quickly. ⚡ Powered by @TokenRouter_US GPU infrastructure 🆓 Limited-time free access 🧪 Ideal for testing, evaluation and experimentation 🏗️ Standard service available for production workloads Note: the free model runs on shared compute, so queues or higher latency may occur during peak demand. For production environments where consistency and stability matter, use the standard GLM-5.3 service. #GLM53 #AI #LLM #Inference #GPU #GenerativeAI #TokenRouter

FREE AI ≠ WEAK AI. I reviewed the current @OpenRouter Free Models collection and ranked the models differently: Not by popularity. Not by parameter count. Not by context window alone. Prioritized: 🧠 Overall intelligence & reasoning 🤖 Agent + tool-use capability 💻 Coding / technical execution 👁️ Multimodal capability 📚 Usable context window ⚡ Speed & efficiency ⚠️ Real-world limits and caveats Current top tier: 🥇 @nvidia Nemotron 3 Ultra — deep reasoning, complex agents & orchestration 🥈 @MiniMax_AI M3 — powerful multimodal + agentic generalist 🥉 Inkling — strong reasoning with very large context But the “best” model depends on the job. For example: → Coding / agents: GLM 5.2, Nemotron, North Mini Code → Multimodal: MiniMax M3, Inkling, Nemotron Nano Omni → Finance: Ling 3.0 Flash Fin → Fast worker agents: Nemotron 3.5 Lightning → Long-context work: MiniMax M3 / Inkling / Nemotron family The bigger lesson: Don’t build an AI agent around one “best model” Build a model portfolio: Powerful reasoning model → difficult decisions Fast model → repetitive execution Specialist model → coding / finance / multimodal Long-context model → research & documents That architecture is especially relevant for platforms such as #Hermes where different worker agents can use different models according to the task. And many of these models currently cost: $0 input + $0 output through @OpenRouter . That makes them extremely useful for experimentation, AI-agent development, training labs and low-cost automation—provided you understand the rate limits and other restrictions. Source: openrouter.ai/collections/fr… Which free model are you using today? #AI #OpenRouter #AIAgents #AgenticAI #LLM #GenerativeAI #OpenSourceAI #Automation #HermesAgent

CAUTION! Free tier models is available for a limited time, may use user interaction and usage data for improvements.

💡 You can use @Meta Muse Spark 1.2 Contributor model (muse-spark-1.2-contributor-free) for free by accessing it through @opencode via their contributor program. meta/muse-spark-1.2-contributor in @OpenRouter openrouter.ai/compare/meta/m…

🇲🇾 Selamat Hari Merdeka! Happy 69th Independence Day, Malaysia! #Merdeka is not only about how far we’ve come — it’s about what we build next. Malaysia’s next chapter will increasingly be shaped by: 🤖 AI 🔐 Cybersecurity 🛡️ Digital resilience ☁️ Cloud 🤝 Stronger technology ecosystems At @arrosoft , we’re proud to contribute to a future where Malaysian businesses can innovate with AI while remaining secure, resilient and ready for change. This Merdeka, one question is worth asking: What can we build today that will make Malaysia stronger tomorrow❓ 🇲🇾 If you’re exploring AI, security or digital resilience, let’s connect and exchange ideas. #Merdeka #Malaysia #AI #CyberSecurity #DigitalResilience

📰 Hermes Agent v0.20.6 (v2026.8.27) — Key Facts github.com/NousResearch/herm… ✅ Major stabilization release with ~525 merged PRs and 1,300+ commits since v0.20.5. ✅ Browser automation can now use the default Chromium profile with user approval. ✅ Stronger remote-agent and SSH fleet management. ✅ Expanded Remote MCP support with 50+ verified vendor-hosted servers. ✅ Faster web workflows through cached web_search and web_extract results. ✅ Improved tool discovery with multi-query search and stemming. ✅ Better long-context efficiency through lean-tail compression. ✅ Optional OS-keychain encryption for stored secrets. ✅ Safer, less disruptive agent updates and better scheduled-task diagnostics. ✅ New model options include GLM-5.3-Flash, MiniMax M3 Free, and MiniMax H3 Max Video. Key takeaway: v0.20.6 is mainly about reliability, remote operations, browser/MCP integration, performance, and security rather than one major headline feature. Have you start your Hermes Ai Agent?

How to get around RM 500 worth of WLD from @worldnetwork 1. Install & setup World App : worldcoin.org/join/642L6FC 2. Verified thru the Orb (now available in KL; TRX & MyEG Tower) 3. Within 24 hours, you will get 25 WLD, worth around RM 200+ and 30 WLD thru 12 months period 🧵