Have you heard these exciting AI news? - February 27, 2026 AI Updates Weekly

Have you heard these exciting AI news? - February 27, 2026 AI Updates Weekly Presented by Lev Selector Slides on GDrive: https://drive.google.com/drive/folders/1dmj00iilSP-JYmriMUUaHCKz5X54e3UX?ths=true Slides on GitHub - https://github.com/lselector/seminar (click on pptx file, then on "raw" or download button on the right) ------------------------------------------------------- Please write in comments your opinion and experience achieving convenience of OpenClaw - but without actually using OpenClaw (using Claude Code of some other tool which is not banned by Anthropic) ------------------------------------------------------- Need AI Consulting ? - Enterprise AI Solutions - https://EAIS.ai - Linkedin - https://www.linkedin.com/in/levselector - GitHub - https://github.com/lselector ------------------------------------------------------- Our books: "Artificial Intelligence In Business: Strategies That Work" https://www.amazon.com/dp/B0FKZ3364G/ "AI Product Manager: Building Tomorrow" https://www.amazon.com/dp/B0FMS9NFK9/ "How to Learn with AI?: A Practical Guide for Students, Professionals, and Self-Learners" https://www.amazon.com/dp/B0FWLNVQMJ/ "Professional + AI: Future-Proofing Your Career" https://www.amazon.com/dp/B0GBXLCTZQ --------- Contents of today's video: 00:00 AI as a teammate you manage 00:27 Harness Engineering is More Important than the Model 00:57 Crowd-sourced "LM Arena" Leaderboard 02:32 Claude Code vs OpenClaw 04:27 You Could've Invented OpenClaw 05:30 Matthew Berman's AI Employee 07:09 Liam Ottley - AI Business Models 08:17 Taalas chips - 17,000 Tok/sec inference 09:25 Chinese AI Firms Distill Anthropic's Claude Models 09:56 ChatGPT upgrades: deep research, code blocks 10:16 Peter Steinberger Workflow 11:33 Example - Create Skills with OpenClaw 12:00 OpenClaw on GPT 5.3 Codex via $20/mo Plus plan 12:42 Matthew Berman's Custom OpenClaw 13:00 OpenClaw + Qwen 3.5 13:32 OpenClaw + Sonnet 4.6 13:58 OpenClaw + Gemini 3.1 Pro 14:19 Oz by Warp - cloud-based AI coding agent platform 14:47 Abacus Deep Agent - Secure 'OpenClaw' 15:31 Abacus AI's "Deep Agent" can create quality audio 15:38 Google restricts Antigravity access for OpenClaw users 16:03 OpenClaw Update (Mistral Chat + Voice + Memory!) 16:21 OpenClaw + Zapier MCP = 8,000 apps 16:38 OpenClaw - Build Mission Control Dashboard 16:51 Three ways to configure OpenClaw 17:31 Make OpenClaw more powerful 18:12 OpenClaw + GLM 4.7 Flash + Gemini 3.1 Pro 18:31 Openclaw: Mission Control + Agent Teams 18:48 OpenClaw - save money, grow your X account 19:08 Run OpenClaw on your Mac Safely 21:00 High-value AI skills to be worth $500K/year by 2027 21:35 Google Lyria 3 AI music model 21:49 IronClaw & ZeroClaw - in Rust 22:10 Why AI generate Garbage Code - and How to Fix It 22:54 India announced three (3) domestically-built models 23:00 VoiceBox - open-source local-first AI voice cloning 23:15 Modeling human concept formation 23:50 Claude Code Agent Modeling Trading 24:10 Google Builds Self-Learning AI (RL2F) 24:34 China leads the humanoid robot revolution 24:53 Claude Coding in COBOL triggered IBM shares price crash 25:50 GGML.ai bought by Hugging Face 26:35 David Silver starts Ineffable Intelligence 27:48 Jobs, Layoffs, demand for AI Engineers ---- Takeaways: 1. Harness engineering around AI models matters more than the models themselves. Models score 90% on benchmarks but only around 20% on real tasks. The difference is how you structure the "harness" around the model. 2. OpenClaw is rapidly growing with 236,000 GitHub stars, but it is mostly a convenience abstraction. Everything it does can also be accomplished with Claude Code and proper prompting. 3. Multi-agent and multi-model architectures are the practical path forward. Companies like AT&T cut costs 90% by matching cheaper, faster models to simpler tasks instead of using one expensive model for everything. ---- Quotes: "AI is no longer a tool you use, but a teammate you manage." "The value is not in the model, but in the skill of the human holding the reins." "OpenClaw is mostly a convenience abstraction, not a fundamental capacity unlock." ---- Summary: The central theme is that harness engineering, meaning how you structure and deploy AI models, matters far more than which model you pick. A large portion of the talk focuses on OpenClaw, the fast-growing open-source AI agent platform now backed by OpenAI, including its integrations with various models like Qwen 3.5, Sonnet 4.6, Gemini 3.1 Pro, and Mistral. Other highlights include Taalas chips running models at 17,000 tokens per second without CPU or GPU, Chinese firms distilling Anthropic models, Claude coding in COBOL causing a 31 billion dollar IBM stock crash, Google Lyria 3 for AI music, and David Silver launching a new startup focused on world models. The talk also touches on layoffs, high-value AI skills, and practical business advice for using AI agents. ---- ---- ----