top of page
BootstrapMD logo.png

Transcript YouTube English auto-captions. ======================================================================== I get hundreds of emails a day, not exaggerating, hundreds every single day from patients, coaches, sponsors, members of my community, people pitching me, people asking for me for things, people copying me on stuff that I don't really need to be copied on. Newsletters I've got I subscribe to, receipts, updates from platforms that I haven't logged into in 8 months. And for years my system for handling all this was and I'm embarrassed to say this, opening the inbox, scrolling, reacting to the one that was the most urgent, the most loud, and then closing it before it ate up the rest of my day. And that was the system, and the joke was there was no system. The system was just me scrolling, hoping nothing important was buried on page four. But a few weeks ago I built something different, and it's not fancy. I'm not running 50 AI tools, I'm not writing code. What I did is I built one AI workflow that takes my inbox, it looks at what's there, and identifies the emails that actually matter, the urgent ones, the ones that I need to act on today, and then it tells me which ones I can ignore. And that's it, that's the whole AI workflow. And the first morning I ran it, instead of getting a wall of 200 emails, I got maybe five to 10. It gave me a short list, and it just clicked for me that not only was I saving time, I was no longer the bottleneck for my own inbox. And that's the shift that I want to talk about today. Because most doctors right now are using AI in the way that that I used to use my inbox reactively. I would just work by doing one question at a time. Maybe ChatGPT, give it a task, give it an email, summarize this article, and that was it. And that's useful, but that's not really where the real leverage of AI is. The real leverage is when you stop using AI as a search box and start using it as a worker with a defined jobs, with instructions, with guardrails. But that's the move from going from ChatGPT to AI agents, and for physician entrepreneurs, it changes everything. My name is Dr. Mike Woo Ming. This is the podcast for physician and healthcare entrepreneurs. This is for folks who want more income, more autonomy, and more control over how they spend their lives. And if you've been listening for a while, you know we've had some really recent episodes on focusing on AI. We've talked about how doctors can use it, how they can use it to to build extra income. We talked about the revolution of AI scribes, the basic AI stack, and why most physicians are using AI wrong. So, today is the next step. I want to go from just using it as a tool to developing workflows. From A asking AI a question to assigning AI a job. Now, the example I gave is I just what built one workflow for my email. I'm not building 50 agents here, but what I am doing is I've identified some workflows that can really streamline what I want to do to build more income for myself and my family. And I'm seeing it inside our own mastermind. We're teaching other students how to do this. I see it with the coachings that I work with. And I want to make sure you, my loyal listener, does don't miss this shift. So, here's what we're going to cover today. We're going to talk about why most physicians are using AI too small. Number two, what an AI agent actually is without all of the hype. Number three, the real reason doctors haven't built one yet, and it's not what you think. Number four, I want to share with you the story of of one of our coaches who's running 11 AI agents across his entire business. Number five, what to delegate first and what to never delegate. And then finally, I want to help you build your first workflow this week. One specific thing that you can do. So, let's get into it. So, why physicians are using AI and it's too small? Because here's a pattern that I'm seeing. I'm seeing with almost every physician who tells me they're quote using AI. They open chat GPT, they type in a question, they get an answer. And they move on. Maybe they'll tell it to write them an email to a patient who missed an appointment. Done. Summarize an article for me. Done. And there's nothing wrong with doing it. And these are real use cases and I do that, too. But it's almost using AI like a vending machine where you insert prompt, receive an answer, and then you move on. But the work still lives in your head. So the next time the same task comes up, you end up starting from back to zero again. So compare that to how you would actually use a let's say a virtual assistant in your practice. You wouldn't ask a virtual assistant a random question every 20 minutes and then send them home. You would give them a role. You'd say, "Your job is to handle these specific things. Here's how I want it done. Here's what to do if you're unsure. Bring me the work, I'll review it, and then we'll go from there." And that's the difference between asking AI a question and assigning AI a job. And the reason most physicians haven't made that jump isn't because they don't understand AI. It's because no one has shown them what a workflow actually looks like in a physician's business. And I want you to think about what you did last week. How many times did you do the same task, the same kind of email, the same kind of summary, same kind of draft, and probably ended up making the same kind of decision over and over again. Every time you do that, every single one of those examples is a candidate for delegation. And not because AI is smart enough to replace you, but because your task already has a pattern. And patterns are what AI is actually good at. So the real leverage isn't building better prompts. The real leverage is building repeatable workflows. Now I want to talk about what an AI agent actually is. And I'll give you the simplest definition that I can. But an AI agent is an AI assigned to completing a defined workflow. Not just answering a single question, but completing a workflow. The workflow has a goal, and instructions, and a source materials, and available constraints, and a review process. If you have a prompt, you might say, "Hey, write me a LinkedIn post about physician entrepreneurship." That's a prompt. Here's what a difference with an agent would be. Agent, a task might be take this podcast transcript, draft a 500 word newsletter, generate five LinkedIn posts in my voice, write three short video hooks, draft a YouTube description, suggest one console called action, and flag anything that needs my review. I don't want you to invent quotes or statistics. If something is unclear, you need to ask me. Now, what I gave you is the same input. It's just me sitting at a keyboard, but you're getting completely different output. The prompt gives you an answer. Agent, what it's giving you is an asset. A supervised agent gives you repeatable leverage. Now, people hear the word agent and it seems it means something autonomous, robots writing emails to your patients while you sleep. That's not what I'm describing. Think of an AI agent as a junior assistant or virtual assistant. It's smart, it's fast, but this time it's tireless. And be honest, sometimes it gets it wrong. Sometimes it's confident when it shouldn't be. But, just as you wouldn't give a junior assistant the keys to your practice on one day, you give them small jobs, see if they do it right, you review the work, and then you would move on. You're building trust slowly. Now, let's talk about why most physicians why they haven't built an AI workflow yet. Usually it it comes down to this. And here a lot of I just don't have time to figure it out. I need to take a course. I'm not technical enough. I'll get to it when things slow down. And I want to gently push back on all of that because I don't think that's the real reason. The reason most physicians haven't built this one yet is they're still using AI to do search instead of using AI to do work. All right? So, I want that mentality to change on here. Because that's the same mentality that we use for Google or searching on PubMed or UpToDate, information in, information out. But, agents aren't search engines. Agents are workers. So, the mental model for worker is completely different. You don't evaluate a worker by whether they gave you a good answer to one question. You evaluated them by whether they completed the job, the draft's ready for your review, organized output, flagged uncertainties, and give you a clean handoff. And I know that part's a little bit hard for physicians cuz we're used to doing everything on there. And it doesn't mean that doing work is allowing you to give control. It means you're letting something else go first, even if it's not perfect, and trusting that your review process to catch what matters. All right. And here's the thing. You guys are already doing this already when you sign up on the scribe note. Maybe your MA is prepping your chart. You're already operating as a reviewer, all right? So, I just want you to do that same shift to instead of you asking for information, you assigning the task. Now, I want to tell you about one of my our coaches, Felix, and I want to show what he's doing cuz it's incredible. He's running 11 AI agents across this entire business. Now, he doesn't have a practice, but he has a practice that involved in in physician in jobs, in job management, in recruiting. And he has 11 defined workflows, each with a job, operating operation agents, content agents who manages his inbox, follow-up agents. And these are workflows that he used to require him to do personally. Or maybe I believe he had a team of contractors doing all this. But now it's run completely using AI agents. He saved real money. He saved lots of time. And that part that gets me, he's also reduced the number of human contractors he's hiring cuz he doesn't need a person to draft every social post and pull every report anymore. He needs a person to review and decide. Now, I'm not telling you need to go build 11 agents this week. All I'm telling you is it starts with one on here. I just want to give you an idea of what the spectrum can be. My first workflow is just I'm just having a simple AI agent build a workflow. And then slowly, I'm having it do operations. I'm having it do marketing. And it's saving me real hours every week, and it makes a real difference. So, let's talk about what to delegate first and what to never delegate. Where should you start? So, my rule is start where AI can reduce friction without creating risk. Low risk, high friction, make it repeatable. All right? Low risk, high friction, repeatable. So, let me tell you how I'm thinking about this in my own business right now, so it's not abstract. So, workflow one was I did email triage, okay? I run it, it saves me hours a week. Workflow two that I'm building is meeting and operations. I'm in a lot of meetings, coaching calls, sponsor calls, internal calls with my team. Every one of those used to leave me with a notebook full of half-legible notes and a vague sense that I was supposed to do something afterwards. Now, I have a workflow that takes a transcript, it pulls out the decision, it lists the action items, it names who owns what, and flags anything that needs a follow-up email I haven't sent yet. It's not perfect, but it gets me where we want to go, and it really improves our next meeting as well because we have the tasks are assigned, it gives them what they need to do, what I need to do, and it just makes our flow a lot better. Right? So, emails and operations, that's where I am, and I think that's where a lot of where most physicians should start. Now, moving on to another agent I'm building is on content repurposing. As you know, I do a podcast, I do webinars, I have newsletters, but those are all being built into social posts, video hooks, descriptions. I'm already doing the hard part, is the intellectual work of actually coming up with a podcast episode. They're doing all of the packaging on there. They're building the website, the blog content, drafting educational content for my sites, frequently asked questions, the kind of stuff patients Google before they ever call. And I'm drafting it with AI, but I'm the one reviewing it, okay? It's helping me build SOPs, standard operating procedures. AI captures the explanations and turns them into a document that can actually be followed instead of me re-explaining it every 6 weeks. Again, these are all examples of low stakes, high friction, and repeatable. This is perfect for workflow territory. Now, what you should never delegate, I think is anything clinical. If it's a patient communication that involves a clinical question, AI drafts it, but it doesn't send it. AI drafts it for me, I read it, I make the changes if I need to, but I'm the one who actually doing it. So, I don't want you to actually auto publish or auto send or auto post on here. There is compliance that needs to be followed, things with marketing claims, anything that could be read as a guarantee. I don't have AI do it. Again, it drafts it, I'm the one who actually posts it, and you need to do the same thing as well. All right. Here's finally I want to talk about what you need to do this week. I want you to pick one task, just one, something that you do every week, you hate doing, and it doesn't require your medical degree. For me, it was email. For you, it might be something different. Maybe it's building a newsletter, meeting summaries, social posts, follows up on consult inquiries, SOPs you keep meaning to write. And then, and this is the part most people skip, is write the job description before you write the prompt. What is the goal of the workflow? What inputs am I going to give it? What outputs do I want? What's the format? What's off limit? What needs to be flagged for my review? So, write that down first, you can dictate it onto your phone or type it onto your Apple notes, whatever. Then go to whatever you like. I use Cloud Co-work. I've also used Manage, but whatever tool you use, if they have agents on it, allow you to create an agent, and I want you to feed that job description plus your source material. Use a safety language somewhere in the prompt. And do not send and I want you to have this in the prompt. Do not send, publish, schedule, or make final recommendations. Draft only. If information is missing or unclear, flag it instead of making assumptions. With that one line that I just gave you, I'll read it again. Do not send, publish, schedule, or make final recommendations. Draft only if information is missing or unclear, flag it instead of making assumptions. That puts you in the reviewer driver's seat. It's the difference between a tool that helps you and a tool that could get you into trouble. And then they want you to run that workflow. And it might take you a few times. See what the output is reliable, make changes, make adjustments on here. Then go ahead and then schedule it. Have that task be scheduled daily, weekly, monthly, whatever it is on there. Again, it is a tool task to help you. It's not just a chatbot. Chatbot was the appetizer. The real meal in AI is delegation. And for physician entrepreneurs, that matters because the bottleneck in your business almost never is intelligence. You're super smart. The bottleneck is time, attention, and execution. And those are the three things AI workflows actually give back to you. You don't have to hand your judgment to AI. Don't. Keep your judgment, stay the decision-maker, but let it prepare the work, let it organize all that chaos, let it draft what you're going to review anyway. This week again, start with one workflow. Mine was email. Pick yours right now. Keep review before send as your safety layer. Build trust with the system slowly. And then once you get that done, build another one. And that's how you move from being the bottleneck to being the grand poobah of your business. Now, if any of this sounds interesting to you and you want to learn more, I highly encourage you to join our school group, Doctors Online Success. We're going to be having a cohort. We're going to be helping physicians, including if I can get him, one of the guys who has 11 agents currently building his practice. And we're looking to help physicians who want to integrate AI into their practice. To find out more, go to Doctors Online Success. There's a a waiting list and we'll let you know when it is available on here. It could be if you need AI they can help you with your practice, your side gig, helping you build your next income stream. Come see us at Doctors Online Success. All right. I hope this was useful for you. Remember, medicine pays the bills, but AI can build that balance sheet for you. I want you to go build something this week. Remember, guys, you're going to have your ups and your downs as a physician entrepreneur. Do something a little each day to get you closer to your goals and keep moving forward.

bottom of page