Hi friends, thank you all for the great community that exists here and to Jake for taking the initiative to set it up. The Backstory (Not essential) I am not a computer or technical worker, but I've been interested in AI driven uses for the past year or so. Initially, I started in Claude Projects. My goal was to make a tool that was full of various context for my life and work, so that the questions I asked and things I was thinking through were always based in my own personal situation. I mainly use this for goal setting and planning. Eventually, it began to hallucinate, and I rebuilt the project, still in Claude, in a different format. It worked well also for a while. The failure points often centered around the release of a new law, and so I figured that there was something going on there because nothing had actually changed in the workings of the project. After that, I shifted to PAI, as it was called at the time, and LifeOS, as it was billed as the exact thing I was looking for. I used and really enjoyed release version 5, but versions 6 and 7 have not been as good to use. I think part of this is due to the shift from Opus model implementation from v4.6 to v5. It also burned a ton of tokens. I had been keeping an eye on ICM and used very simple structures and systems of my own for local projects. The LifeOS project itself was a more complicated version of ICM, or at least a tangent derived from it. The Question Many of the examples and use cases for these ICM structures that are discussed here are focused on producing actual work and deliverables. Does anyone have an example of how they have used it or seen it used for a personal information and planning system? The Followup How much context is actually necessary for good output to be received? For example, in a build for a customer, how much detail about the company and the work they are doing goes into the setup? How much of that is relevant for the ICM to reference and understand as it is working through the material outputs? In my head, I've been told and understood that having a better background helped produce better results, but context windows can only be so large. Where is the trade-off between the AI having no outside context, some outside context, or lots of outside context as it relates to output quality?

This is happening to me all the time now, and either I wasn't catching it before, or something changed. I do realize that this might be operator error, which is why I'm asking for help. Have you been able to wrangle the "hallucinations" which let's just call them what they are, fiction and/or lies. One request: please don't blindly copy/paste outputs if you run this post with AI. I'm not saying don't do that, just politely asking if you could give it a read and if the output makes sense to you, drop it in the comments. Here are two examples: --------------- Example #1 --------------- Had Claude draft this up, and in the process it created a rule for it without me asking. What's frustrating about this is that now I have to read through another markdown file to ensure accuracy and consistency to how I do things, and I still have doubt that it'll even honor the rule it wrote for itself. The admission: 1. What I did. Credited three voice rules to NAME, and wrote that he flagged them in a comment on your Voiceprint LinkedIn post. The comment doesn't exist. I checked his comments feed, his posts, Gmail, Slack, and Skool. The earliest trace is my own /session capture on 8/29, with nothing upstream of it. I can't reconstruct where the name came from, and I didn't invent a story for how I invented it. 2. It happened twice. Yesterday I told you the attribution had reached the public repo. It hadn't. I repeated the session log's "added all three to..." line without opening the file. That second one is the more useful failure, because it shows the actual mechanism: an unsourced sentence in a workspace file gets read as a source by whoever comes next, including me. 3. Six guardrails were loaded and all six missed. constraints.md bans invented attributions by name. Its final scan says every claim traces to a source. Shared rule 0 forces that file to load every EOD run. Root map rule 5 says verify the live system. Digirol bans "I think" and demands a real search before calling a contact new.

Do not ban your staff from using AI. If you do, this is what will quietly happen. They'll go into stealth mode. They'll still use it, but this time they'll get better at hiding it. They'll strip off every AI fingerprint, watermark and metadata. You'd think your 'No AI policy' worked, but really you just empowered your workforce with unregulated firepower. And frankly, would you blame them? Would you have done anything differently? Imagine being asked to take an hour-long walk to the office every day when you have a vehicle that can do it in 10 minutes. And when asked, management says it's for fitness purposes. Right thinking, wrong application. Here's what to do instead and it's in two parts: Part A - Internal work 1. Take responsibility Understand that your people are not saboteurs; they are practical. This helps you approach the situation not only with empathy but also practicality. They found tools that make their week lighter, and in the absence of direction from above, they supplied their own. 2. Seek AI Clarity Many times we've seen leaders shut down beneficial operations because they didn't understand them and, therefore, couldn't govern them. AI Clarity is not about learning how to use AI tools. It is understanding what it is, what it can do and the risk (P&L) associated with it so that you can make informed decisions without bias. Part B - External work While you gain clarity, here's what you can do from day 1. 3. Bring the usage to light. Announce, without threat, that you want to know how people are actually using AI today. Amnesty first, direction second. You cannot govern what your people are hiding from you, and they will hide it precisely as long as the official posture is denial. 4. Start with the lowest hanging fruits You do not need a complete framework to name the non-negotiables. Client data never enters unapproved tools. Nothing AI-drafted leaves the building without a named human owner. Decisions about people are made by people. Publishing three enforceable lines this month beats publishing thirty aspirational ones next year, and it signals that direction has arrived.

Hi everyone, For those who have used both Claude Code and Codex through their subscription plans, is Codex actually better and faster for vibe coding? I have been using Claude Code on a subscription for a while. It is often very capable, but I have found it extremely slow. I switched from the Mac desktop app to VS Code, hoping it would improve the experience, but many tasks still take a very long time. Lately, I have also noticed more mistakes and patches that frequently break things. I am now spending more time debugging and repairing its changes than making progress on my projects. I have tried several Claude models, including Fable 5, Opus 5, Opus 4.8, and now Fable 5.1. I have heard that Codex is faster and more efficient, but I would appreciate clarification from people who have used both: 1. Can Codex be used through a subscription, such as the $200 ChatGPT plan, with coding limits comparable to Claude Code? Or does extensive use require separate API billing? 2. Is Codex genuinely faster and more reliable than Claude Code for vibe coding? 3. Can Codex take over projects that were started in Claude Code by reading the existing codebase, `CLAUDE.md`, and other Claude project files? I would especially appreciate feedback from anyone who has used both platforms extensively on real projects. Thank you so much for your help.🙏

Here is a slightly odd question. I have recently upgraded to premium to get the most of this community, however, I'm finding myself still doing the same thing I did on the free! That obviously reflect on me rather than the opportunities available here. Could anyone, On premium describe how they maximised the opportunities available in the community when they upgraded to premium? For example, how should I use discord? I have installed it? But not sure what to write in it. Do i ask questions there? what is a good way of using it? How do I make best use of the templates that come with the premium subscription?