Question for those using Claude Code / Codex with real client work: How do you think about confidential client material when using a personal Claude or OpenAI account? I’m trying to understand the practical standard for responsible handling of client data — especially when working with client files in a local folder structure through personal Claude Code / Codex. Is this normal responsible AI-assisted work if settings are configured correctly, or should confidential client material only be handled through business/enterprise setups, signed DPAs, or local/self-hosted environments? Especially interested in how people advising serious clients / regulated industries are handling this in practice.

EDUBA @Kay K has had the challenging task to redesigning my Company website to fit the folder and Unix theme. It may seem odd at first but as you scroll down through the site you will see the influence of traditional and fundamental structures, the simplicity of it is what I love. Figured I would share this here and share his hard work as I love the idea of making a website folder based, because as you know, I love my folders. EDUBA

Claude Mythos just quietly changed the conversation around AI and most people are not paying attention yet. This is not another small upgrade. Anthropic is calling it a “step change” in capability and placing it in a completely new tier above their existing models. That alone should tell you something is different. What makes Mythos stand out is not just intelligence but autonomy. You give it a simple goal like “find a vulnerability” and it does the rest. It reads code, forms hypotheses, tests them in a live environment, and even writes working exploits on its own. No hand holding. In testing, it discovered thousands of zero day vulnerabilities across major systems and browsers. Some of these bugs survived decades of human review and millions of automated tests. Mythos found them anyway. It can also chain multiple vulnerabilities together to escalate attacks, something that normally requires highly skilled human experts. And it does this repeatedly with a high success rate. This is where things get interesting. The same system that can secure software at an unprecedented level can also be used to break it. That dual use nature is why models like this are not being fully released yet. We are moving from AI as a tool to AI as an active operator. Not just assisting humans but independently exploring, testing, and executing complex tasks. If you are building in tech, cybersecurity, or AI, this is not something to ignore. This is a preview of what the next generation of systems will look like. And it raises a simple question Are we ready for AI that does not just suggest but acts?

Been quiet for a reason. I got flown out to New York to work on a project with a major artist. Grammy-level work. Can't say who yet (NDA), but if you know your way around Wu-Tang or Def Jam circles, you might be able to guess. I've been in the studio every day working on the album with them, building out the tech side of something that hasn't been done before in the music industry. Tomorrow's VIP High Tea session I'm breaking down how this deal came together, what the tech stack looks like, and how AI is opening doors in entertainment that didn't exist a year ago. I'll also be answering questions from the Google form submissions. If you've been on the fence about VIP, this is the session to show up for. See you tomorrow ✌️

This is a "God Tier" stack for anyone building agentic workflows or highly automated development environments. - per Ai. - This is NOT everything but its a large chunk.- ENOY! 🧠 Category 1: Orchestration & Swarm Frameworks The "Brains" that coordinate multiple agents into a single functional unit. 1. Antigravity-Awesome-Skills - URL: https://github.com/cleodin/antigravity-awesome-skills - Need: To move beyond a single chatbot into a "Skill-Based" autonomous system. - Want: Access to 250+ pre-configured expert skills (n8n, SQL, React) for your agents. - Why: It allows the agent to "level up" its capabilities on the fly based on the task at hand. 2. Microsoft Agent-Lightning - URL: https://github.com/microsoft/agent-lightning - Need: High-speed, low-latency agentic responses for real-time IDE interaction. - Want: To eliminate the "lag" typical of standard LLM chain frameworks. - Why: It optimizes the interaction loop, making the AI feel like a snappy, native part of the OS. 3. OpenAI Swarms - URL: https://github.com/openai/swarm - Need: A lightweight, educational framework for multi-agent handoffs. - Want: To test how a "Triage Agent" delegates work to specialized sub-agents. - Why: It’s the baseline protocol for how different AI personalities talk to one another. 4. Pydantic-AI - URL: https://ai.pydantic.dev/ (Main Repo: https://github.com/pydantic/pydantic-ai) - Need: Strict type-safety. You cannot have an AI returning "text" when your app needs "JSON." - Why: It forces the LLM output into a rigid schema, preventing "Unexpected Token" crashes. 5. Agency-Agents (AionUI) - URL: https://github.com/msitarzewski/agency-agents - Need: Specialized personas (Designers, QA, Accountants) instead of a generic assistant. - Want: 110+ structured AI personas to improve output quality by up to 70%. - Why: It provides the "Expertise" layer that ensures your code follows senior-level standards.