Peace & blessings AI family, I know I haven’t posted much (because I’ve been building), but I would like to get your ass thoughts on my website and offer. It’s not fully together yet and yes, there are some things to change from an architecture perspective and graphics or screenshots to be added, but I’m putting my website out there to get immediate feedback. Thank you again for your help and support https://ahvance.com/index.html -Robert Runyon

Hello all, I’m trying to get to level 3 to improve my existing AIOS product I’m building for my clients. I want to incorporate the aspect of an autonomous agent/wizard that holds your hand through the set up (accomplishes two things: familiarity with the agent, thorough onboarding without the time consuming manual interview process). This same agent then becomes your AIOS assistant, teaching you how to use the system, updating skills for you (this is what autonomous agents excel at) and offering a text first interface for your AIOS interactions. Ask the agent to run skills, create scheduled skills, pull data from connected tools, etc. I’m working on a prototype now for a few clients and for myself. would love feedback and help getting to level 3 to see what areas I can improve based on the AIOS video. Thanks in advance and happy to connect. Current stage:

Original post: https://www.skool.com/ai-automation-society/agency-owner-building-aios-as-a-service?p=4de9cf74 A while back I posted about building an AIOS as a service for my marketing agency clients, and about wanting an autonomous agent that holds your hand through setup and then becomes your ongoing assistant. Quick progress report: The agent does both jobs now. - Setup: it runs a conversational onboarding that builds your folder structure and business context for you. No manual interview, and you get comfortable with the agent from the first message because you watch it work. - Day to day: the same agent is your AIOS assistant. It teaches you the system lesson by lesson, builds and updates your skills, runs them on demand or on a schedule, and pulls from your connected tools. Everything is text-first, so you just ask. A few design choices that paid off: - Everything lives in plain files you download and own, not a vendor's cloud or a vector DB. It's portable, and any model can reason over it. (I am combining ICM, OKF and LLM-Wiki) - Capabilities are skills you build and run, not a fixed roster of agents. Much easier to grow and maintain. - It's provider-agnostic, so you can point it at whatever model you want and it isn't locked to one AI provider. - I added a simple dashboard you can preview before you stand it up, so you see your command center before committing to it. - I also ended up taking it past agencies. It now adapts to a few other kinds of local and online business, which was a nice validation that the method generalizes. Right now it's a working prototype I'm rolling out to a couple of clients and using it to run my agency, which is awesome. I'd really value feedback from this group on where the gaps are and what would push it further. Happy to connect, compare notes, or give anyone a walkthrough. Current stage:

Anthropic just dropped Claude Fable-5, a Mythos-class model that sits above Opus and is available to everyone right now, plus Claude Mythos Five for Project Glasswing partners only. I break down what the two models actually are, the pricing, why Fable-5 is only free on your plan until June 22nd, and the benchmark jumps in coding, knowledge work, and vision. I also revisit my earlier Mythos prediction and what I'm most excited to put Fable-5 to work on.

Hi everyone. I am curious if anyone here has built or worked on something similar. I’m exploring how AI and automation can be used with a CRM and/or client portal where external users have to authenticate before interacting with data or business processes. The general use case would be users who are completely outside of the organization logging into some type of portal, app, or front-end experience and then interacting with CRM-related workflows across several activities. I’d love to hear from anyone who has done work in this space, especially around: How did you approach the overall architecture? How did you handle authentication for external users? I’d appreciate any examples, lessons learned, or “watch out for this” advice from people who have been down this road. Thanks in advance!