I've been building Rex and Piper, two AI agents that help with sales prospecting for my higher-ed SaaS business. Here's a simplified version of how I structure Rex's instructions: - Identity: Act as a B2B researcher specializing in higher education. - Task: Research a college and determine whether it's a good prospect for our offering. - Context: Use our ideal customer profile, product information, and qualification criteria. - Constraints: Use verifiable sources, don't make assumptions, and identify missing information. - Output Format: Provide a score, supporting evidence, and a Pursue/Review/Hold/Do Not Pursue recommendation. Each part has a purpose. Identity establishes the expertise, Task defines the job, Context provides the business knowledge, Constraints help prevent bad assumptions, and Output Format makes results consistent. I expected Claude to give me better-qualified prospects rather than generic research, with fewer revisions needed. My biggest takeaway is that I don't necessarily need to repeat all five parts in every prompt. With ICM, much of the Identity, Context, and Constraints can live in reusable files that Claude already knows how to navigate. That's what I'm working toward with my Second Brain: better results, less repeated explanation, and context I can reuse across different AI tools.

I've been building an AI Workspace using the same principles as the three-file system, but expanded to support multiple projects and AI tools. My instruction files define how AI should work with me. My project files describe what I'm building, the goals, and what success looks like. My reference files provide supporting knowledge, documentation, and examples. One example is my RentalManager project. Instead of repeatedly explaining the business requirements, workflows, and rules, Claude can navigate the workspace and use that context when answering questions or making changes. The biggest benefit is having a single source of truth that works across Claude, ChatGPT, and other AI tools. Key takeaway: Better-organized context means less repeated explanation, more consistent results, and easier collaboration across AI models.

The biggest takeaway for me was how a simple folder structure and a few instruction files can replace a lot of complex AI infrastructure. Seeing how an agent can navigate folders and find the context it needs really made it click. I've already started building my own AI Workspace using this approach. My goal is to maintain one source of truth that I can share across ChatGPT, Claude, and other AI tools. Going forward, I want to keep improving that structure so I can switch between models, compare results, and avoid duplicating context.

I'm using AI across several business and personal projects and moving from exploration into building structured systems. What caught my attention about ICM is creating reusable context that works across multiple AI tools. This makes it easier to compare results without starting from scratch each time. For my start up, I'm building AI "employees": Rex, who researches and qualifies potential customers, and Piper, who uses that research to prepare personalized sales outreach for my review. They aren't great yet but are slowing improving. Next up is a marketing role (not sure what name I'll give it yet, maybe Marvin). One great take away is that I don't need these to be separate agents, just different roles in the same structure. I'm excited to keep learning how to build practical AI systems that save time, reduce complexity, and help grow my businesses. Happy building everyone!

Hey Clief Notes fam! 👋 I'm Scott, and I'm excited to be here. 🙋 A little about me: I'm a longtime tech entrepreneur and former CTO. After selling my software company a few years ago, I'm getting back into building businesses, this time with a strong focus on AI. Outside of work, I'm a private pilot, avid traveler, and aspiring content creator. 🎯 My current goal: I want to move beyond just experimenting with AI tools and learn how to put AI to work in real-world applications — building businesses, improving productivity, and creating new opportunities. 💪 What I'm currently building/working on: I'm involved in a SaaS startup serving higher education and am building an AI consulting practice to help local small businesses adopt AI and other technology. I'm also interested in using AI to build a social media presence around my aviation adventures. 🤔 My biggest struggle or question right now: There's so much happening in AI that it's hard to know where to focus. I'm trying to figure out which tools and skills are worth investing time in, how to move from experimenting to actually implementing useful solutions, and how to avoid chasing every shiny new thing. Looking forward to learning from everyone and sharing what I figure out along the way! 🚀