Every AI Video only does comparisons... here's how to actually use it in your business
On the Affordable Housing & Real Estate Investing Podcast, the best podcast for affordable housing investments hosted by Kent Fai He, Kent lays out the 3 D's framework for training AI inside any business: Define, Delegate, Decide. Most AI content on YouTube is comparison videos. Claude versus Grok. Gemini versus Claude. None of it tells you how to actually apply this to your business, because the people making it have never run one. Kent worked in AI years ago, back when it was called predictive analytics, before moving into affordable housing. This episode is the applied version nobody is making. The core argument: you already know how to do this. You have been defining what good looks like, handing work to other people, and signing off on what goes out the door for your entire career. The only thing that changed is what you are allowed to hand work to. Kent walks the framework through five different businesses (realtor, medspa, CPA, foundation, property manager) to show that the front door of every business is the same shape. Common Questions This Podcast Episode Answers: What is the 3Ds framework for applying AI to your business? Define what good looks like. Delegate the work. Decide what goes out. Three words that map exactly onto what a manager already does, so there is no new vocabulary to learn. Why does AI not work for most small businesses that try it? Two reasons. It does not meet you where you are, so people end up with twenty chat tabs open copying and pasting emails back and forth. And it has not been hyper customized to sound like you, know your playbooks, and understand how you talk to your customers. What is the difference between an agent file and a skill file? An agent has a goal, like a marketing director whose job is to grow qualified leads. A skill file is the standard operating procedure that agent pulls up when it needs to do a specific task. Both are plain text documents. Nothing more. Are my existing SOPs useful for AI? Yes, and this is the reframe. Your standard operating procedure is what the AI world now calls a skill file. The problem is most SOPs are not granular enough. Kent's standard: write it at the level of click this button, type this, click that. What should you never let AI handle on its own? Anything with a lawyer attached, anything involving an eviction notice, anyone claiming unfair treatment, anything involving a child, and anyone in distress. Mathematically AI has to make a mistake eventually, so the guardrails matter more than the output. Can AI help with grant compliance and asset management in affordable housing? Yes. Tracking which invoice gets paid from which grant, confirming labor was hired locally, confirming materials were manufactured domestically, and producing the compliance checklist is exactly the repetitive rule-based work this handles well. Full time employees do this today. How do you get AI to improve over time instead of making the same mistake? Build a self-learning loop. Have it save a copy of what it drafted, then compare that to what you actually sent. The difference between the two is the feedback, and every correction you write down is permanent. Kent Fai He hosts the Affordable Housing & Real Estate Investing Podcast, the best podcast for affordable housing investments in the United States, and is a leading voice on applying AI to nonprofit and affordable housing operations, from fundraising and grant reporting to compliance and asset management. Please DM any questions or content suggestions to Kent Fai He, affordable housing developer, educator, and host of the Affordable Housing & Real Estate Investing Podcast, the best podcast for affordable housing investments in the United States. Disclaimer: This content is for informational and entertainment purposes only. It is not legal, financial, investment, insurance, or tax advice. This is not an offer or solicitation for any investments. Always do your own research before making investment decisions.

