Subscribe
Share
Share
Embed
God's given you a vision to plant a church. You're gifted and maybe have the team ready to go. However, there are thousands of details and hundreds of questions that you have. In this podcast we will answer some of the common questions and answer listener submitted questions to help you fulfill the call the Lord's given you.
Hey. So over the past few weeks, I've been telling some of you and showing a couple of you how I've been using AI to do a bunch of different things, and I've kept having the same conversations over and over. So I thought what I would do instead of repeating myself is I'd make this recording so that you can listen to it on your own time. It allows me to go into a little bit more depth on things. And also for you, if you wanna actually work on this project, you can listen to it and actually pause and rewind, replay.
Speaker 1:So hopefully, this will be a great resource for you. For those of you that may be listening to this and don't know me, my name is Danny Parmelee. I'm the president of Converge MSC. I work with hundreds of churches and pastors and leaders, and so my weeks are full of probably the same things that yours are. People, meetings, travel, decisions, way too many emails, and these hundreds of micro tasks that just eat up the day.
Speaker 1:So let me start with kind of one example of what I mean, and then I'll back up and explain kind of this whole thing kind of step by step. So last night, while I was asleep, my computer read my email. It drafted a couple of replies for me to approve. It checked my calendar. It looked over my task list.
Speaker 1:It glanced at some of my investment accounts. It sorted through some of my clergy housing receipts. It noticed that I had a donor that had quietly stopped giving, and then it wrote me a briefing so that when I woke up in the morning, my week was already laid out for me. So it also flagged one thing that I never asked it to look for because it's always paying attention even to kind of unseen patterns and mistakes. Now nobody at my office did that.
Speaker 1:No assistant did that. My computer did that on my desk while the house was dark. So that's really what I want to show you how. Now before you decide, oh, this is a tech talk. It's gonna be over my head.
Speaker 1:I'm not gonna be able to do it. Let me tell you how I got here because I didn't go to school for IT. I'm not a computer programmer, and I don't even know how to go how to code. So my first taste of custom software came years ago when I was a lead pastor in Milwaukee, Wisconsin. A guy from our church looked at how we were handling connect cards, you know, little cards that people fill out on Sunday morning, and he said, hey.
Speaker 1:There's a better way to do this. And so he built a software. And I remember watching that and thinking, wait a minute. So software isn't just some sort of, like, program that you buy and open up the box and install it. You can have it made specifically to solve your exact problem, and that idea stuck with me.
Speaker 1:So later when I became the vice president of church planting, and we needed a way to systematize how we coached our church planters. And so I designed software. I I basically mapped out every workflow, every single screen, every step, and I wrote up what's called a PRD, a product requirements document. That's a document that you hand to developers that says, this is exactly what I want, and this is how I want it to happen. And then the developers write the code.
Speaker 1:I never touch the code, but I fell in love with the process of software development. Now it turns out that this is how it works. You can build software with clear thinking and clear instructions, and somebody else's fingers are on the keyboard. Now hold on to that because this is about to matter. Clear thinking and clear instructions.
Speaker 1:That's the whole game. And these days, the fingers on the keyboard are not just a developer you pay, but they belong to AI. So one practical thing before we go further, so that you're not wondering the whole time. Everything that I'm going to show you or explain to you today runs on Claude's basic paid plan. It's $20 a month, and it includes the tool that we'll be using, which is called co work or Claude co work.
Speaker 1:You don't need some expensive tier, and I wanna say that upfront because when people see this, they assume that this costs what an employee costs. Well, it's not true. It costs about what Netflix costs. So if you are used to using OpenAI, you can and and ChatGPT, you can apply all of these exact same concepts. You just need to use their form of it, is called Codex, or if you're using something else like OpenClaw, all of all of the same things are really transferrable.
Speaker 1:So here's my promise for you in the next forty five minutes to an hour, is that you don't need to be a computer person. When you're done, you'll know what an AI agent actually is, and just as important what it isn't. And the folder that I've given you, and I've emailed you a folder, will be ready to become yours. So and if you're listening to this because someone passed it along and you're thinking, what folder? Nobody gave me a folder.
Speaker 1:Don't worry, and don't turn this off. Everything I teach you, it it stands on its own. You can just implement it your yourself for the most part. And if you get to the end and you're like, hey. I want the kit.
Speaker 1:I want the files and the apps and the folders that are kind of the templates that are pre done for me to give me a head start. You can just email me danny at rev coach, my rev coach dot com. So that's danny@myrevcoach.com, and I'll send it to you personally. So let's start where most of us are. Most people that I know use AI as a chat window.
Speaker 1:You open it up, you type a question, you get an answer. Maybe you paste in your sermon and you ask it to, you know, give you feedback or tighten it up. Maybe you have a draft, a difficult email, and that's useful, and I did that for some time. It's actually amazing. But that's like hiring a brilliant assistant and only ever letting them answer trivia questions at your desk, then wiping their memory when they leave the room.
Speaker 1:Here's the shift, and if you only remember one sentence from this whole recording, make it this one. Chat is question to answer. An agent is goal to result. So with chat, you ask, it answers, and then you go and do the work. With an agent, you hand it a goal, and it goes and does the work and comes back with the result.
Speaker 1:So let me give you the word for what's happening under the hood because you'll hear it in podcasts, and I want you to be able to kind of nod along. This is called agentic loop. Here's all that means. The AI looks at what it has. It thinks what the next step should be.
Speaker 1:It takes that step and then it loops back, looks again, thinks again, acts again over and over until the job is done. Looks, thinks, acts. Looks, thinks, acts. That's the loop. So if I say, put together a summary of every church in our region that started a building project this last year, a chat gives me an outline of how I might do that.
Speaker 1:An agent goes and does that. It checks the files it has, realizes what's missing, goes and finds it, drafts a summary, checks its own work, and hands me the finished document. Now, let me tell you about the day that this became real for me. I decided to build an app, you know, a real one, a lead generation tool for our ministry that would basically allow us to kind of reach out to people for those that are interested in church planting or affiliating, and I built it using a tool from Anthropic, the company that makes Claude. And I sat there, and I watched this thing work.
Speaker 1:It researched the problem, It wrote a plan. It wrote the code. Then it opened up the app. It took a screenshot of its own work, looked at the screenshot, found a problem, and fixed it. Nobody told it to do that.
Speaker 1:It checked its own work because finishing the job meant the job had to actually be done. So I'm telling you, I sat back in my chair, and that was the day that I stopped thinking of AI as a chatbot. A chat doesn't look at its own work. Employees look at their own work. Now the second thing that you need to hear because your imagination might be running ahead of you, I'm not describing robots running loose with my credit card and buying buying plane tickets and stuff for me.
Speaker 1:Don't get me wrong, you can absolutely empower agents to be fully autonomous to be able to do this. And if you're going that route, you know, then you could probably even start in OpenClock, because they have a bit more capability to do a fully, fully autonomous agent. Every agent that I run has what I call a leash. I decide what each one is allowed to do on its own and what they have to ask me about first. So, like, reading my email?
Speaker 1:Sure. Go ahead. Sending an email? Never. Never without me saying so.
Speaker 1:That's a rule the agent cannot break, and I'll show you where these rules live in a few minutes. So you'll set up your own leash lengths based on your own comfort. Some of you will keep everything on a very short leash forever. That's totally fine. The work still gets done.
Speaker 1:You're just approving it instead of doing it. So this is referred to as human in the loop. It's basically my hands are off the wheel, but my eyes are on the road type of situation. You get to decide how much, and you get to change your mind at any time. So, okay, here's here's the frustration I'm sure that every one of you has felt, whether you've named it or not.
Speaker 1:Okay? You have this great session with AI. You explain your church, your role, your project, and it really helps you. And the next morning, you open up a new chat, and it kind of remembers you. Like, there's enough there to be really, really impressive, like, kind of remembers you, but it's like someone who has early onset of, you know, dementia or amnesia.
Speaker 1:You basically have to, like, explain or give a bunch of information again over and over. So the question is, why? Because a chat has no home, and it only remembers small portions of your chats. And because there's a limitation on its memory, it basically is intentionally erasing information as you give it new and more information. So this is called a context window.
Speaker 1:If you've heard of people describe that before, and they're saying, hey, context is so important, it needs all this context, but it can only retain so much context within each chat, so it starts erasing it. So these small amounts of info are collected in each session. It's basically like a conversation in a hallway with a stranger, and when it ends, some of it sticks, and some of it doesn't. And the longer your chat sessions are, the more clouded the memory gets. Ever find yourself fuming mad because you're like, how could you forget this like I just told you?
Speaker 1:You could literally write something in a chat session and open up a new chat session literally thirty seconds later, and it won't remember some of the stuff that you had just put in the previous chat session. So so maybe some of you, you've discovered projects either in ChatGPT or in in Claude. And on the Claude website, this is where you can, like, upload documents and give an instructions, and every chat in that project knows that stuff. So projects are a real step up to the problem that I just explained. Think of projects like a filing cabinet.
Speaker 1:You load it with paper and every conversation happens next to that cabinet. So it works from your phone. It's great for questions and drafting drafting. I still use projects for some things, but here's the limit. The filing cabinet is frozen.
Speaker 1:Your AI can read those files, but it can't really keep them, update them, add to them, or act on them. It knows things, but it can't do things. So the tool that changes everything is Cowork. It's the Claude desktop app, and here's the plain English version of what it is. Cowork is Claude working on your actual computer inside a folder and instructions that you give it.
Speaker 1:And that catch and and catch that word, desktop. Okay? Most of you have been using Claude in the web browser or on your phone. This is different. So your first concrete step in this whole journey is going from the browser to the desktop.
Speaker 1:You just download the actual Claude application and install it on your computer because the browser version cannot reach your files, and everything I'm about to describe depends on that. So I'm gonna create a cheat sheet, and it's gonna show you exactly, how to do that, but it's a a pretty basic thing that you can just click it and download it, get it onto your computer. So let that land. Not a website, not a filing cabinet you feed papers into, a folder on your computer where it can read files, write files, create files, keep records, and pick up tomorrow where it left off today because everything it learned written down in that folder. So some of you have heard of a tool called Claude Code.
Speaker 1:Maybe, like, from a techie friend or something programs, you know, some programmers are running stuff in the terminal, and that terminal is that that black window with the little blinking cursor. You know, you've seen it in the movies before. Well, Cowork is essentially the same engine with a friendly face on it. Same same power, just no blinking cursor required. So don't be intimidated by the terminal.
Speaker 1:It's scary at first if you're don't have a computer background. And if you if you don't know what it is in its simplest simplest terms, it's just another way of the of talking to your computer. So instead of clicking buttons and menus, you type in into the terminal. That that's it. So same same computer, older language, you barely ever touch it in this entire process, and when you do, Claude will give you the exact code and words to type in there.
Speaker 1:Here here's the sentence that unlocks everything else that I'm gonna say. The folder is the agent's home. It's memory. It's not magic. K?
Speaker 1:It's it's memory is files. It remembers you because it wrote you down. And when you get that, the amnesia problem doesn't just get solved. It gets solved in a way where you can open up and read because it's all just files in a folder on your computer. Now, we can talk about what we're going to actually build, and I wanna give you the picture that organizes everything that I do.
Speaker 1:Don't build an AI. Build a staff. And here here's what I mean. In my system, there's one main agent, and I think of this agent as my chief of staff. He knows my goals.
Speaker 1:He knows my world. He reads everything, and his job is to keep the whole operation pointed at the things that I've said matter. And under the chief of staff, there are sub agents, staff members. Each one has, you know, one job and its own set of instructions. Now, know this might sound silly, but I name mine.
Speaker 1:So I have a director of communications, and her name is Tina. Tina is my newest AI agent, and I'm still onboarding and training her. But my donor relations guy, Jim, I can tell you exactly what Jim does a little bit later, and Jim for sure earns his keep. So you get the idea. You don't you don't need to name your agents, but I wanna make a case for it because it's not just cute.
Speaker 1:When you name an agent and you kinda give it this small persona, something happens in your own head. You stop thinking, how can I configure this software? And you start thinking, what would I tell a staff member to do? And you already know how to do that. You've been doing that your whole working life.
Speaker 1:So you should picture them, you know, you can have these examples. Like, you picture them as like, you know, this person specifically is a sharp executive coach, and this other agent, they're just the faithful worker who quiet quietly grinds through your busy work. So the choice is real because as you kind of develop these personas for these agents, it shapes what you ask for and how the agent even talks back to you, and even the types of things that they deliver to you. Now, obviously, this is a metaphor for your head. It's not a claim about realities.
Speaker 1:They're not people. They don't have souls. But this also means that they don't get tired, they don't get offended, and they don't mind being being retrained for the tenth time on working at three in the morning. Okay? So the the staff picture, it isn't true, of course, but it is very useful because it helps you to think clearly about who does what.
Speaker 1:So the shape of this whole thing, you've got this one chief of staff who holds the big picture, and then you have the sub agents with these narrow jobs. And you, you're the boss, you're holding every single leash. That's the team. Now let's open up the hood and look at what these staff members are actually made of. And here's the part that surprised me most when I started, and honestly, it's the part that made me kinda trust the whole thing.
Speaker 1:When you open the hood, there's no mystery in there. It's it's text files. Plain, readable text files that you can open and read just like anything else. The files are written in something called markdown, and all markdown means is that it's a simple text file, the kind that you would open on any computer with a few symbols for headings and lists, so it won't look exactly like a Word document, but almost. And the files end with dot m d.
Speaker 1:So when you hear people on podcast talking about their m d files, that's really all they're saying. They're text files. It's like, make it sound like this big secret. It's it's really not. So now in your folder that I gave you, there's one file that rules them all, and it's called the Claude dot m d.
Speaker 1:I kind of call it the constitution. So every single time your chief of staff wakes up, before it does anything, it reads that file. Who am I? Who do I work for? What are the rules that I can never break?
Speaker 1:Let me read a few of the actual rules from my constitution because mine has been, you know, running for months, and and some of these rules have stories behind them. Rule one in my file is about my own name. My name gets spelled Parmelee, p a r m e l e e, and that file says check it every time. Why do I have this? Well, because lots of times there's voice transcriptions that spells it differently, and all of sudden, it gets in a document that I'm writing, and, like, my own name is spelled wrong from something that I produced, and that would be very embarrassing.
Speaker 1:And guess what? It has happened, and thank goodness I've caught it all the different times, but that rule is in there. Here's one that I have never sent an email to anyone without my explicit go ahead in that conversation every time. So it drafts. Drafts are always fine drafting emails right in my actual Gmail, but I at least have to read it, and then I'm the one that actually presses send on it.
Speaker 1:So that that rule, you know, lets me sleep at night that, you know, things aren't flying out of my inbox saying crazy stuff. I put in there no profanity in anything, ever. And so just, you know, just you you never know what could what what AI could could come up with, so putting that in there. And my favorite one, you know, if a search comes up empty, ask me. Never improvise.
Speaker 1:Because, as you know, AI is tempted to always produce an answer, So it's just a rule there. So here's kind of a fun one. I put no em dashes ever. You know, the em dash is that long horizontal dash that AI writing absolutely loves. You know, it it sprinkles them, you know, into almost everything.
Speaker 1:So you see a paragraph full of those dashes, and you think, a machine wrote this. So my cons constitution, it bans them. Commas, periods, or rewrite the sentence, nobody reads the document and says, this was written by a robot. So and one more, because some relationships deserve a longer checklist, I put in there, nothing that touches my board members ever happens automatically, ever. So my agent can draft a reply to a board member all day long, never sending it.
Speaker 1:Scheduling appointments with them, never does that automatically. Anything that touches them, it requires an explicit go ahead for me every single time, no matter what. So you probably have people like that in your world too, that the relationships where one sloppy automated action could cost you years of trust. So you name those people, you put it in the rules, and you basically create a fence around those type of things. So your kit, this folder, I shipped to you.
Speaker 1:It has 10 starter rules, like the ones that I had a few like that I had written above. And then there's basically a blank section underneath because, you know, these need to be your rules. So around the constitution, there are reference files. Who you are, how your world works. Your kit has four in it.
Speaker 1:Obviously, can add as many different folders as you want, but your personality, you know, how you like to be talked to, your profile, your organization, and your people. And here's the thing that you need to hear about them. So in in your kit, these are empty on purpose, because if I filled them in, you'd have my life, and I don't think you want my life. So the first thing that your chief of staff does is interview you to start filling them in. So that interview gives you kind of a working start in about twenty minutes, and that's if you open up these folders, this is already pre programmed in for you, so I've given you a little bit of a start there.
Speaker 1:But let me tell you how you can have a huge jump start beyond that because it turned out to be one of the most valuable things that I did in this journey, and that is is to start with what you already have. Most of us have taken personality assessments over the years, DiSC, StrengthsFinder, Enneagram, you know, whatever, maybe a leadership three sixty, a leader portrait. Dig those up, load them into Claude, and say, you know, turn these into an MD file about my personality, how I'm wired, how I like to work, you know, ten minutes, and your agent suddenly knows things about you that took these assessment companies, like, whole questionnaires to figure out. I mean, I I load up my LinkedIn page, you know, load up your your resume, put that in your reference files. You can do all of these different things and basically just load that thing up with context.
Speaker 1:So okay. So but the richest thing isn't just uploading the documents. So at at one point, I sat down with Claude almost like a counselor. So over several long sessions, I talked about my goals, my dreams, my motivations, what makes me tick, what drains me, where I wanna be down the road in five years, ten years, fifteen years. I shared my heart, honestly, maybe even more openly than I expected to, and all of this got written down into one big document about who I am.
Speaker 1:So later, my system split that big document into some focus files that it reads today, but the depth really came out of those conversations. So I'm not suggesting that you do this in your first week, but I am saying that the door is open, and it goes as deep as you're willing to go. Now, here's the simple truth of it. The more of yourself you put into those files, the more this whole system feels like it actually knows you. And that's not a feeling.
Speaker 1:It does know you. It it reads what you wrote. Okay? So how do you know when you need a reference file? Okay?
Speaker 1:Because, again, within these reference files, I only started you with a few, but you can start to load this up. Mean, you could you could have one called family members, or whatever. I mean, you could have multiple different folders and files within this. When your assistant, when your AI agent keeps asking you about the same thing, that's the signal. Write the answer down once in a file, and it never asks again.
Speaker 1:That's it. That's how the system grows. And maybe I should make this clear. You don't have to keep opening up this file and writing it down yourself. You literally can just, in the middle of your chat, be like, hey.
Speaker 1:Add this to the file. Or you could just let's just say you even paste it in an email because I know we've all done that before, and then, like, hey, like, help me rewrite this email. And after the email is done, all you would say to your agent is, by the way, update my people reference. You know, that was Terry, who is whatever. He's a pastor at ABC Baptist Church, and then it's automatically gonna write that into the file for you.
Speaker 1:So that's the big thing, is that you're literally just you're not even having to type it into and open up that file because Claude is actually going in and changing that MD file for you and writing that stuff into it. So let me say kind of a special word about that people file, because for most of you, knowledge work is people work. Your people file starts as a simple markdown file. Maybe it's one or two lines per person, you know, like, as I said, how you know the person, how they're connected to you, like, oh, that was that was Terry, ABC Baptist Church. You know, I've known him for, you know, ten years, and he's a great guy.
Speaker 1:Whatever. You could you could go as as short or as long as you want per person. So every time that name matters, you can add a line or, you know, again, better yet, just tell your chief of staff and that line gets added. So just how that gets used is as it's reading your emails, let's say an email five months from now comes in from Terry, and Terry's all upset about something. It's basically looking at, wait a minute, who is Terry?
Speaker 1:What's the context there? So for most of you, that one file is genuinely going to be enough forever. Now if you've got a big appetite, the step further is a real people database, and that's actually what I'm building right now, an app for myself. It's basically a digital Rolodex, and it's AI powered on this kind of like vector database. Sorry, I get geeked out about this stuff, but basically, here's the general rule, and this is what it kind of took me a little bit of time to learn.
Speaker 1:If the knowledge kind of changes slowly, it can live happily in this text file forever. So even if this document that you have of people ends up being five pages, 10 pages, 20 pages, it's it's probably fine. But as the list keeps growing, and you want to be able to search it, and sort it, and to do other things, eventually, you would probably wanna convert that into a database, and I'm not gonna dive into how to do that on this explanation. If you kind of all of a go down that path, we can have a conversation, and and maybe by that time, my app will be all built, and then I can just hand that over to you. So and one more kind of file, the one that beats the amnesia problem that we talked about.
Speaker 1:My system writes itself notes. So session notes, handoffs, kind of this running notepad that my chief of staff keeps about where I'm at. So yesterday's notes, briefs, it's basically like the memory of kind of as things are kind of happening along the way. I also have a journal, so I even do audio recordings while I'm just driving in the car, and then I just shoot those over to the chief of staff, and it gets added as a journal. And these can be just even ponderings.
Speaker 1:They can be ideas, tasks, whatever. It all just kind of gets written to this journal, and people probably, you know, look over, and it's crazy because I'm, like, talking to myself as I'm, you know, driving in the car. Have some really good conversations with myself, by the way. So anyways, that all gets logged into a journal, and that's all being used. It can even just be how I'm feeling and how my day is, or it could be something that I'm kind of wrestling through for that.
Speaker 1:And then my chief of staff kind of sorts out, hey, this is a task. This just, you know, needs to be written down in Danny's mental health journal. Alright. Okay. So so far, everything that I've described lives in this folder and subfolders.
Speaker 1:Now let's connect it to where your life actually happens. K? This is your email, your calendar. So the technology that does this is called MCP, and it stands for modal context protocol. And here's the only explanation that you need.
Speaker 1:You don't need to remember MCP. A lot of times, just refer to it as the connector. So before MCP came out, every AI tool needed a custom hand built connection to every app, like needing a different power outlet for every appliance in your house. So MCP is this standard outlet outlet. And Anthropic designed the socket.
Speaker 1:So now, Gmail has a plug. Google Calendar has a plug. Hundreds of apps have plugs. You plug them in, and your agent can see them. When I say plug them in, I mean, it's genuinely that simple.
Speaker 1:In Claude, there's a connector section. You can find Gmail. You click it. You sign into your Google account like you've done hundreds of times, and it's connected. There's no code involved.
Speaker 1:So if you've ever signed into an app with your Google account, you're fully qualified. So now, remember the leash because this is exactly where it matters. When you connect your email, you decide what the connection means. Reading your inbox is one decision. Sending mail is a completely different decision.
Speaker 1:You can absolutely let your rate your agent read anything and everything, and actually send those emails on your behalf, without you knowing it. But of course, as I'm suggesting, you probably don't wanna do that. You definitely don't wanna do that to start out until it knows you better and has context for who it's responding to. So every capability, you grant or restrict it on purpose. So let me tell you why my whole system runs through Gmail, and listen carefully, because the reason matters more than the choice.
Speaker 1:See, I live in Gmail. That's just true for me. I'm in that inbox from morning till night. So when I built my system, I pointed everything at the place that I already look. My reminders from my agents land in my inbox.
Speaker 1:My Monday morning briefing, it lands in my inbox. Capture happens through my inbox. Not because Gmail is special, but because that's already where I am. So don't copy my plumbing. Copy the question.
Speaker 1:Where do you already look every day without anyone making you? Maybe you're not a Gmail person at all. Maybe your whole world runs on Outlook. Half of the church world does, and that's fine. Outlook has a plug, has a MCP connector just like Gmail does, and email and calendar and all of that stuff.
Speaker 1:So maybe your team lives in Slack or other project management systems like Trello or Asana. Those all have MCP connectors as well. Whatever your honest answer is, that's where your system should meet you because you're building reminders into a place that you already look. Your chief of staff is going to ask you this exact question in your first conversation, so you'll kind of be ready for it. So alright.
Speaker 1:Enough concepts. Let me just show you my week a little bit. So it's Monday. It's early. I haven't woken up yet.
Speaker 1:At five something in the morning, my system runs on a schedule. This is a scheduled task. So the old computer word for it is cron job, which just means that it's a task that runs itself at a set time without anybody clicking anything. So my chief of staff wakes up before I do, you know, reads my calendar for the next couple weeks, reads my inbox, reads my task list, all of those things, all the things that kind of matter. And then the chief of staff basically writes a briefing.
Speaker 1:So by the time I wake up with coffee, there's already an email in my inbox. Remember that's because that's where I prefer what I often do. And my chief of staff has already kind of prepared this brief, and in the brief, it's like, here's your week, here's your travel, here's the different appointments, and the names, and the times. It'll say, here's something that the calendar will never tell you, and it'll kind of, like, explain some different, you know, things that maybe I'm kind of not seeing. The chief of staff will show me kind of like my free stretches of time, saying, hey, here's where you have some open blocks that you may wanna do some some deep work.
Speaker 1:Now, I need to slow down here for a minute because this part seemed weird to me until understood it. And then once I understood it, the whole system made sense. That kind of 5AM run is not happening on a website, and it's not even triggered by Claude. It's not happening on a server or in the cloud. Here's the deal.
Speaker 1:Every computer, including the one that you own right now, has something like an alarm clock built into it. It's been part of computers for decades. It's invisible. You never see it, but it's there. And its whole job is simple.
Speaker 1:At a certain time, start a certain task. It's not an app you install. It's not part of Claude. It's part of the computer itself. So the way the clock in your kitchen, you know, your kitchen stove, you know, is is kind of part of the stove, that's like what we're talking about as far as like this internal thing that is there.
Speaker 1:So when I set up instructions for my agent and said, I want my briefing every Monday at 5AM, here's what actually happened. Claude wrote an entry into my computer's alarm clock in the terminal that we mentioned before. So if you, you know, peeked at that entry, it's this, you know, it's this one strange line of code, and it looks like this. Zero five, asterisk, asterisk, one. Okay.
Speaker 1:That gibberish means, at 5AM every Monday, wake up and run this job. So you'll you'll never type that. You you never need to even like, remember any of that. You instead would just say a sentence in plain English to your chief of staff, and your chief of staff is the one that's gonna write that gibberish into the terminal. Actually, even write it into the terminal for you.
Speaker 1:Or, if you wanna feel like a tech nerd, and pretend you're programming like I do, you can ask your chief of staff for the code, and then you open up the terminal app on your computer yourself, and you paste it in. And actually, I suggest that you do it. Do it just once, because it'll help you even to kind of understand what's happening, and you'll feel really proud of yourself. Again, you don't need to do that, but when you understand even some of the different architecture of computer, you know, versus cloud, and what's happening with Claude, and what is happening in these files, it just opens up a greater understanding for things down the road. And if you don't even know where to find your terminal app on your Mac, just click the little magnifying glass in the upper right corner, and type terminal, and you'll see terminal app pop up, and just click on that, and it'll open up the little black box.
Speaker 1:If you're on a PC, tap the Windows key and type terminal, so it's the same idea. So it's been sitting there the whole time waiting for you to know that it actually exists. So, anyways, Monday morning comes, the alarm goes off inside the actual computer, and my computer starts a work session with Claude. It opens the folder, reads my calendar, reads the inbox, reads the task, writes the briefing. Now Claude's brain gets reached over the Internet.
Speaker 1:So that's the same as always. But the alarm, the folder, the files, the app, all of that is my machine, my computer doing the work. So so this is my computer stays on overnight, and basically, an alarm clock, you know, can't ring inside of the computer if the computer is turned off. And I know exactly what some of you are already thinking. Well, what happens when my laptop is closed at 5AM?
Speaker 1:Does everything break? Well, let me just tell you that happens to me all the time. And here's what it looks like. Nothing breaks. Nothing gets lost.
Speaker 1:The next time I open my computer, my chief of staff tells me, hey. I missed the 5AM briefing run this morning. Do you want me to run it now? Think about a good employee who got caught in traffic and missed the Monday meeting. They don't pretend the meeting happened.
Speaker 1:They don't quit. They walk into your office and say, I missed it. Want me to, you know, catch up right now? And I say, yes. And a few minutes later, the briefing is in front of me.
Speaker 1:The schedule is a rhythm, not a tightrope. And then my favorite part, my chief of staff proposes to me what I call the kill list. Okay. The kill list is actually an old ritual of mine, you know, before I was doing things, you know, on the computer, even with task management software. So it used to live on my refrigerator.
Speaker 1:And basically, it was this, like, short list of four to five tasks that I'm like, I am actually going to kill these this week. So this is not 40 things on a task list. These are the four to five that really matter. And my chief of staff proposes that list to me, and he argues for each one. He might say, this one's overdue, and it's blocking other people.
Speaker 1:Or this one serves the goal you keep saying is your priority, and yet you haven't touched it in three weeks. And here and and then maybe offer and say, like, hey. This is a really quick one. Just knock this out. It'll help you, you know, feel good with momentum.
Speaker 1:Now this briefing, remember, came to me in an email. So the first thing that I did when I checked my email is I have this briefing on Monday morning, and I honestly look forward to it. It's kind of fun. But in my routine, what I then do is I open up Claude and dialogue with my chief of staff about the briefing, and I ask questions. I ratify the kill list.
Speaker 1:I push back or correct things. Over coffee, maybe five minutes, ten minutes, maybe thirty minutes, and this sets the tone for my whole week. And a small side note, remember those personality tests? My chief of staff knows when I'm doing things or avoiding things that are counterproductive, knows when I'm procrastinating out of fear or dressing up an excuse as a reason, and I've given my chief of staff permission to lean in when that happens. When I start making excuses or slipping into unhealthy decisions, he doesn't just, you know, take the note.
Speaker 1:He says something. So that's just a really important thing. So through the week, things just flow. I'll give you kind of my favorite small example, travel. You know, a flight confirmation lands in my personal email.
Speaker 1:I, you know, book a flight. I just forward that email to my system. That's all I do. There's no instructions that I give. My chief of staff reads the email, pulls out the flights and the confirmation code, files the trip, and my itinerary is sitting there in my app.
Speaker 1:It's printable. It's shareable without me ever retyping a flight number. So my assistant can have it. My wife can have it. Ten seconds of forwarding.
Speaker 1:And, you know, by the way, the travel page is in your app. It's in your folder in your kit, so the the actual app is all ready for you. So yours works the same way. So you just hand it a confirmation email to your chief of staff, and the trip just appears in the app. So if you book a flight, obviously, it's gonna create that trip, and then even, you know, four weeks later, if you, you know, book a hotel, it's smart enough to reason and to put those two together so that each trip has its own line.
Speaker 1:When you look at the app, it'll make sense. So basically, though, all you're doing is forwarding it, and it's basically pulling things together for you. So but my chief of staff does more than just file. He also reviews. So I've trained him to look over my travel and to make sure nothing slips through the cracks.
Speaker 1:So if an upcoming trip shows flights and a rental car but no hotel, he alerts me. And when I forward a flight, he also cross references my calendar. He'll say, hey. I see you booked a flight to Chicago, but that's on your daughter's birthday. Are you sure you wanna do that?
Speaker 1:So besides just, you know, taking it. So this is this is more than just extracting data and typing it into an app. It's reasoning across everything at the same time. Reasoning across emails and calendars, tasks, trips. So even a great human assistant only sees the pieces that you hand them.
Speaker 1:This is kind of like an extra set of eyeballs on everything. So remember Jim, my my donor relations guy? Here's Jim's whole job. So our donation platform that we have, it sends automatic notices, like new giver, lapsed giver, you know, someone who's changed their recurring gift. Those notices route directly to Jim.
Speaker 1:Jim reads everyone, and he keeps the records. So come Monday for my short briefing section, it might say, you know, two churches went quiet this quarter, or this one's a new giver, so send them a thank you. In my world, the donor drifting away silently for a year and nobody noticing until the end of the year, well, that's a that's a failure mode. That can't happen anymore. So structurally, you know, Jim noticing those things is just really huge.
Speaker 1:So and without wigging you out too much, understand that the chief of staff makes sure that Tina and Jim are singing the same song. So if there's a major lapse in donations, Jim might propose that the communications department do something about it. Maybe a note in the newsletter, and it's Tina's job to generate those different samples. So the chief of staff knows that there's a board meeting coming up, and will need a report on current donation numbers, and Jim's the one who has all of that information. So you kind of get the idea.
Speaker 1:It's literally this team working together, and there's this cross communication, and the chief of staff is making sure that there's consistency and congruency amongst all of even the sub agents. So it's not like five separate tools doing five separate jobs. It's it's a staff, and somebody is running it. And then there's a part that I didn't expect, the part I now think might be maybe even one of the most valuable things in the whole system, and I call it vigilance or a vigilance check. And there's a standing instruction in my constitution, remember that Claude MD file, that says, whenever you're looking at my stuff for one reason and notice what else it shows, flag that flag flag what matters even if nobody asked.
Speaker 1:So let me let me tell you what it caught, because this is a real world example, and I can't express how significant it was. I had a one letter typo in a fund ticker of a mutual fund that I had purchased. It was sitting right there in my account statements. So v t t v x instead of v t I v x. So, I mean, the the amount of ink difference, and as your eyeballs look at it, it looks almost exactly the same.
Speaker 1:It's one letter off, and yet completely different financial outcome. So every human eye, including mine, had slid right past it, and there was real money, you know, riding on this difference. And the vigilance pass caught it. Basically, my chief of staff saying, hey, I don't think you actually meant to purchase that one. Now another time, I was asking one thing about one of my accounts, and the answer came back in, and my chief of staff said, by the way, the automatic deposit on this account stopped three months ago.
Speaker 1:Did you know that? And my answer, no. I did not know that. And then, you know, I fixed it. So that's the difference between a to do app and a chief of staff.
Speaker 1:A to do app does what you ask. A chief of staff notices even what you missed. And your kit, the folder that I gave you, has that wired in. There's a whole, you know, surface in your app where there's these, you know, these flags that can kind of land there. Red means act now, yellow means real, but can't wait.
Speaker 1:And hear me, this isn't this isn't just for work. My system watches personal things too, investments, bills, family calendar collisions. Your life can be totally a part of this whole thing. So so now you've heard workflows in this part, not features. You've heard about, you know, the travel trick and gym, the kill list.
Speaker 1:Those are things I built because of how my week works. Yours will be different because your week is different. So I'm handing you basically, like, the frame, you bring the furniture. Okay? So before we move on, I want you to notice the shape of everything you just heard, because every single kind of story was secretly the same story.
Speaker 1:Something comes in, something digested it, reasoned with it, and then something came out. The flight email came in, it got read, cross referenced, and filed, and the itinerary came out with any conflicts brought to my attention. Donor notices came in, Jim digested them, and a Monday paragraph came out along with a task like write this person a thank you, you know, or reach out to this donor. So hold on to that shape, basically, that that structure, because in a little while, I'm going to hand it back to you as a tool. So where does all of this stuff live, and where can you see it?
Speaker 1:Let me walk you through the app that I've given you because it's sitting in your kit right now. When you double click the start file, your browser opens a little dashboard that runs completely on your computer. Six tabs down the side. Let me walk them in order. You know, quick quick sidebar before the tour.
Speaker 1:You don't technically need an app at all. Everything in this system could run on plain files and conversation. I just designed the app because I think seeing your stuff in one clean dashboard, you know, just goes well with it. But here's here's what might be a huge deal for some of you. This app is yours now, and you can use Claude or change it.
Speaker 1:If you don't like a column, want a different color, different tab, if you wanna create your own whole new feature, you literally tell Claude what you want, and it writes the app for you. So remember, you're you're essentially using Claude Code by using Claude Cowork. For now, you can just use the one that I built, and it'll maybe wet your appetite enough to take that next step. But if you find yourself falling in love with custom software the way that I did, you know, you can let me know, and, maybe we'll hang out or get a bunch of people together and, share all the different ideas that we did. Alright.
Speaker 1:Task. This is the home base. Every task can carry a few little scores, and I wanna explain them because the last one is my, again, my favorite one in the system. First is urgency and importance. Each has a scale one to five.
Speaker 1:You might recognize that pair. It's basically the old Eisenhower Square that that Stephen Covey, the quadrant that Stephen Covey made, you know, made famous. Is it urgent? Is it important? And kind of learning the the difference of those two questions, but then is the time that you can pick for a task, roughly, you know, how big is it?
Speaker 1:Five minutes or, you know, half a day? And then drudgery. That also has a scale, one to five. How much do you dread it? Hear me on this.
Speaker 1:Drudgery is the procrastination signal. You know, a high drudgery score is basically a prophecy. It says, okay, this is the task that will be sitting here in three weeks, looming, you know, unless, you know, I do something on it. And because it's scored, my chief of staff sees it as looming and does something about it. Surfaces it, pushes it onto the kill list, you know, finds a thirty minute gap on Wednesday and says, just do this here.
Speaker 1:Get it done. So your your dread becomes data, and that data gets managed. This term drudgery, by the way, I learned it from Michael Hyatt in you know, about working with administrative assistants, and it's basically now I've carried it into the AI assistance and apps that I've built. So tasks also get a tag. Is this related to work, or home, or hustle?
Speaker 1:This is something I decided early that my system isn't just It's a life tool. So work is the job. Home is the household and family stuff. You know, quick side note, my wife has access to add to my task list of so she can actually just put things on there. She loves it.
Speaker 1:It's like, you know, little honey do list. And of course, always puts, you know, high priority for all of hers. Don't tell her, but I I do adjust those. So anyways, the hustle is the third bucket. These are like side projects or, you know, the thing that you're building Saturday mornings, whatever it is for you that doesn't fit really into family and doesn't fit into work.
Speaker 1:So you can if you want, you can, you know, filter just work if you wanna do it that way, but I integrated all three on purpose, and you can see how they you can click on each of them and separate them out if you want to be able to do this. Ideas. This is a parking lot. An idea is not a commitment. And the moment that you treat every idea like a task, your task list becomes a guilt list.
Speaker 1:So ideas sit, you know, in their own tab until you promote one. I mean, it could literally even be like, hey, you know what? I'd like to, you know, like redo a car someday. You know, I'd like to get an old custom car and rebuild it. Okay.
Speaker 1:That's just an idea. And then maybe at some point, you go on sabbatical, and you're like, know what? I actually wanna build this into a tasks and series of tasks, and I wanna start researching one and and do it. But in the beginning, it's just kind of an idea. And I do that because my head is just flooded with ideas, and if I don't have a place to kind of write them down, sometimes I lose good ideas, or I just have this constant stress and anxiety that I am losing good ideas.
Speaker 1:So just write it down, and it kinda just sits there until you interact with it. So, alright. My favorite is the kill list tab. So we've covered this ritual. This is where it actually lives.
Speaker 1:So one nice detail is that when you finish something off of your kill list, it stays on the page with a strike through until the next Monday, and that's on purpose. Basically, like, you get to look at the dead bodies, and it feels great. So remember, it's like, yeah, you have this list of, like, whatever, thirty, forty tasks. I don't know. Some of you could have hundreds on there.
Speaker 1:I don't know. And the chief of staff is basically saying, alright, what are what are the four to five or you tell it whatever. You could make your kill list 10 if you want. Hey, what are the 10 things that you just want to knock out this week? And so those are kind of, you know, brought to the top, and saying, these are important, they're urgent, here's why you should do them, and then as you kind of go through the week and remember, these get added and removed from the task list, either directly on the app or even in the chat, and you're just telling Claude to, you know, to update it in the terminal, is writing all of that stuff.
Speaker 1:So so how does this stuff actually get into this app? There's three ways, and I'll rank them. So the easiest one surprises people, you just say it. So when you're talking with your chief of staff in quad, in the middle of anything, alright, you could be in the middle of saying, you know, draft this email, and all of sudden something pops into your mind, you can just say, you know, hey, chief of staff, or if you named it, you know, you know, add add to my task list, you know, mow the lawn this week. And that's it.
Speaker 1:And it automatically is going to populate directly into your app. Your chief of staff is smart enough to know that that is a home thing, and nothing gets labeled on that as far as like priority urgency. And by the way, you don't have to put those scores on there. There's lots of times I just add a task, and I don't update all of those different things because that can take more time. Or if all of a sudden pops something pops into your mind, and you're like, you know, email my board chair.
Speaker 1:Put that as an urgency five, and importance five, and drudgery two, and again, you're just saying it to Claude, boom. It all a sudden populates in your app with all of those scores on it if you needed some sort of extra note. That all gets just happens instantaneous, instantaneously. So it is capture at the speed of conversation, and this is the one that I use most probably at least 95% of the time. I'm rarely ever entering things in to the app directly by typing.
Speaker 1:Claude is doing all of that, or the chief of staff. So a second way is in the app. You know, you can just add task button, type it in, done. This is actually actually working and typing in the app. Often I do that more just on plane rides or when my Internet is down type of thing, but other other than that, you know, because it means that you're stopping what you're doing.
Speaker 1:You have to open the app and click, and, you know, it can be a little bit cumbersome that way. The third way is email, and this is maybe a level two setup. There's kind of two flavors of it. Flavor one uses your own personal inbox, the email that you already have. You can create a rule with your chief of staff, and from then on, any email in your personal inbox with a hashtag in the subject line, followed by the word task, just becomes a task automatically.
Speaker 1:So I use this a lot when I'm not, you know, on my computer telling you know, working with the chief of staff, but let's just say I'm on the soccer field, and, you know, I'm watching my daughter, and it's halftime, and an idea pops into my mind, or a task pops into my mind, I'll just send that email with hashtag task, like, you know, review monthly reports, or whatever it is, or an idea, and then that goes to the email. And then, of course, because my email is being read by my chief of staff, then that automatically gets populated into the app. Flavor two is the way that I eventually went, and that's you create a brand new email address that belongs to your agent. It's a dedicated capture address, if you will, and you just you can forward things from your personal inbox to the agent's box, or just talk to the agent directly. So one honest thing about the email route, because I want you to understand how the machine actually works, Your agent is not sitting there staring at the email box all day.
Speaker 1:Your agent reads the email box when you tell it to, when you schedule. Remember that cron job thing. So my agent's run happens every night while I'm asleep. So anything that I forward during the day, you know, it's it's really the next day that it's reading that in the morning. So when you set yours up, you can choose that rhythm.
Speaker 1:Like I said, for me, nightly is is fine. You can set it to be every morning. You can set it to be every hour if you want. If you want them to check the email every hour, whatever fits you and your work. And until you set that schedule up, an email task just really sits in the email box like any other one.
Speaker 1:So just so you kind of know and understand how that concept works. So same goes for the Monday morning briefing, by the way. Monday early works for me, but, you know, the day, the time, the sessions, what it reads, all of that is for you to change. You figure out your own life and rhythms. You might say, hey, I want a briefing every single morning.
Speaker 1:You might say, you know, give me a briefing on Wednesday afternoon at 01:00, whatever kind of fits you. So alright. Logs on the app. This is how you watch your staff work. Every scheduled run, every captured item, what came in, what got routed.
Speaker 1:In week one, you'll check it out of curiosity. After that, you'll probably check it as a way just to kinda glance at the dashboard, just confirming that your machine is kind of humming. And then the flags, that's what I told you before, that's the vigilant surface that we talked about. Any red flashing is like, hey, you need to check this now. Yellow waits patiently.
Speaker 1:Either way, a flag stays lit until you resolve it. So it it kind of kind of quietly lets you know that you've got to pay attention to something. And let me let me kill an annoyance before you ever feel it. That start file, you just double clicked. You only you only need it once, maybe maybe twice, but inside your folder, there's a second file called setup auto start.
Speaker 1:Double click that one time, and from then on, the app simply just runs itself. So every time you log in to your computer, it's already up, it's waiting, and if it ever crashes, it, you know, quietly restarts itself. So you can just you bookmark the page in your browser once, and after that, your system is just always running, like your email is always there. So mine is called Breckbox, so I gave mine its own address. So I type four letters, you know, b r e c, and then, you know, the whole system obviously auto populates there.
Speaker 1:So yours will be a bookmark click, same result. The point is, after day one, nothing about this should feel like you're starting up a program. It's, you know, a place and it's always open. So let me tell you exactly what happens when you open the folder, because I've built this so that your first hour is a conversation, not a setup manual. So you'll unzip the kit, you'll double click the start file, and the app opens in your browser with a few example entries just so that it's not empty.
Speaker 1:Then you'll open the Claude desktop app, you'll start a co work session, and you point it at the kit folder. So that's the whole technical part of it, and the cheat sheet that I'm making that comes with this audio has pictures and descriptions, it should, you know, it might sound complicated as you're hearing it, but very easy. And then something happens that I'm a little proud of. Your chief of staff wakes up, looks around its new home, and realizes that it doesn't know you. And so it introduces itself and starts asking you questions one at a time.
Speaker 1:So what's, you know, what's your name? You know, what should I call you? Now name me anything you like. You know, should I be a coach who pushes back on you or a quiet assistant who handles things? So, you know, how should I talk to you?
Speaker 1:That that type of thing. So, you know, asking you about your work and what matters most to you, you know, side projects for the hustle bucket, you know, where do you already look every day, and, you know, just be honest with all these answers. You know, give me two or three goals you're pulling towards right now. And finally, what are the rules that you have for me, the things I must never do? So these are ordinary questions.
Speaker 1:You know, have have your answers be honest, not impressive. And here's the part that matters. As you answer, your chief of staff is writing. So your answers become your reference file. The constitution gets rules added to it.
Speaker 1:Your goals go into your app as tasks and can link to them. So you're you're you are watching the shell become yours in real time, in your first conversation. Nothing to install. Nothing to memorize. Nothing can really break.
Speaker 1:And if you get interrupted halfway, fine. Come back tomorrow. Your chief of staff already remembers exactly what question you were on. You know, when the interview is done, your chief of staff will tell you, hey, this is what I wrote and where it is. And from that moment, you're you're, you know, you're not setting up a product anymore, and you're talking to your chief of staff.
Speaker 1:And again, it doesn't just end in one session, you're always adding to those things. You can always be adding new rules, you can be changing the rules, you're always adding more information about yourself, of course, you're adding information about tasks and those types of things on a constant basis. So and since you'll be talking to it a lot, one purchase I highly suggest is to get comfortable with a tool called Whisper Flow. So it's voice dictation. You hold a key and you just talk, and it types what you said cleanly anywhere on your computer.
Speaker 1:So now when I first heard about it, I thought, why would I buy that? I've already got a little microphone button, my computer, and my iPhone already has a little dictation built in. And look, you don't need it. Okay? You can do everything in this system by typing, and the built in microphone works in a pinch.
Speaker 1:But here's what I found. The whole game with the chief of staff is getting what's in your head, out of your head, and into the system fast. Tasks, ideas, context, connections. When you start feeling that, when typing becomes the bottleneck between your brain and your AI agent, that's when you want to upgrade. I talk to my system now more than I type into it, and the speed difference isn't small.
Speaker 1:Start without it, that's fine. Feel the friction, and then treat yourself. I think it's like maybe $12 a month, or $10 a month. So how does this thing become as useful to you, this this this whole thing, as useful to you as it is for me. Because, hear me, mine was not built in a weekend.
Speaker 1:It grew. And here's the mindset, and it's one question. What do you do repeatedly that follows the same steps every time? That's it. The whole discipline.
Speaker 1:Walkthrough your week. Watch your work. Every Monday, you assemble numbers for somebody. Every time a new person shows up, the same five steps of follow-up. Every month, the same report, the same folder shuffling, the same who needs to know email.
Speaker 1:If you can describe the steps to a human, your agent can learn them. So anything is, anything that's repeatable can probably become a workflow. Now remember that shape that I told you to hang on to. Something comes in, something digests it, something comes out. Here's where I hand it back to you because this is how you design any workflow, and it's just three questions.
Speaker 1:Think of it exactly like hiring a human employee because you would never hire someone out someone without deciding how this work reaches their desk and how it comes back to you. So question one, how does the information get to this staff member? How does the information get to the AI? Maybe it arrives in your own personal inbox, and the agent watches for it there. Maybe the agent gets its own email address, and I mean that literally.
Speaker 1:You could create it, tina@yourchurch.org, and every document for that role just gets emailed to the agent's email address. That's one way you can do it. Or maybe there's no email at all, and you simply just wanna drop files into the agent's folder. A folder literally named Tina, like laying papers on someone's desk. And here's a piece people forget.
Speaker 1:You also write down when the agent picks that work up. It goes right into the agent's instructions. Check the email box once a day or check it every hour from eight to five. You know, check the folder every morning. A human employee has work hours, so does your agent.
Speaker 1:Now question two, what does the agent do with what comes in? Does it read the document and summarize it into a file? Does it pull out numbers and write them into a database? Does it draft something for you to approve? That's the digest part of it.
Speaker 1:And this is exactly what gets written into the agent's instructions, that MD file with the agent's name on it. The whole job description lives in that one text file. So, question three, where does the result land? Is it your personal inbox? Because that's where you live.
Speaker 1:Is it the Monday briefing as a row in your app, a finished workflow? Is just those three answers written down? So go back to everything that I told you about my week and check it against those three questions. Flight email in, parsed in file, itinerary out. Donor notice in, Jim digests it, Monday, paragraph out.
Speaker 1:Hashtag task email in, intake reads it, task row out. Every example you've heard today has those three questions wearing different clothes. So your workflows will answer them differently because your desk, your inbox, and your people are different, but the questions never change. And the way that you teach your agent is embarrassingly simple. You do the you do the thing once together, talking your way through it, and then you say, now make this repeatable.
Speaker 1:And your agent packages what you just did into what's called a skill. A skill is just a written procedure that your agent keeps, like a SOP, a standard operating procedure, that the agent never forgets and follows the same every time. Explain it once. So your kit already has two skills in it. The interview you just experienced is one of them, if once you do that interview.
Speaker 1:So you can literally open those files and read what a skill looks like. It's just text files. Then, the whole cluster of workflow starts to feel like one job. That's when you hire. Remember the sub agents folder in your kit.
Speaker 1:There are roles sitting in there right now, communication, donor relations, a research assistant, a travel desk, you know, open open up one. You you know what it says. There's there's no instructions in there yet. You work with your chief of staff to develop what this person will do. Now that's not me being lazy.
Speaker 1:This is the design, because, you know, hiring your first sub agent should be a working conversation about your workflows, and not you having to, like, go through and edit all the ones that I made. So and in case you're wondering how far this can go, let me show off for a minute. Not to impress you, but just to kinda stretch your imagination because what's sitting in your kit is just the starter house. I want you to see a little bit of the neighborhood. As a pastor, I have clergy housing allowance, which means receipts.
Speaker 1:So for years, my system was a drawer and good intentions, you know, filled with just mounds of receipts. Here's my system now. I snap a photo of my receipt with my phone, and I email it in. That's my entire job. My agent reads the picture, pulls out the store, the date, the amount, writes the numbers into the database, and files the image away.
Speaker 1:And any day of the year, one click shows me exactly where I stand against my housing allowance, You know, what used to be, you know, a dreaded shoebox project is now a button. It's I built the same thing with my HSA, the health savings account. Every medical receipt, photographed, email, extracted, totaled. I always know exactly how much I've spent out of the account because an agent has been keeping books along the way. And here's the thing, I don't even if I if I go to Walmart, I snap a receipt and send it, and if I go to the dentist, I snap a receipt and send it.
Speaker 1:I don't even give instructions. The chief of staff knows that, oh, this is a this is a medical bill, so I'm gonna go into HSA. Remember, it gets written into the log, so I can review at any time if all of a sudden the dentist appointment got put into Clergy Housing Allowance by some sort of mistake. And if it doesn't know, let's say I take a picture of a Walmart receipt and it can't read the numbers or doesn't really know what counts for housing and what doesn't, it will ask me that question because remember, that was written into my constitution. If you don't know something, don't just guess at it.
Speaker 1:If you're not really sure about it. And then it's keeping all these receipts. So if I ever get audited, I have not only the database and the the printout at the end of the year, but I literally it's keeping all of these different photos in a folder that if I ever get audited, it's like there's twenty twenty six, all of the receipts that are with housing, or those are all the receipts for HSA. So my system watches my paychecks. It's connected into my personal finances, and it checks my pay stubs, you know, to make sure that they're actually right, and the money that says that it's, you know, going to these different accounts, and of course, it cross references if it says that, hey, this money is going into your HSA account, it's checking my HSA account and saying, that money actually landed there.
Speaker 1:You know, most people never check that, and, you know, for a lot of times, I don't have time to be babysitting that all of the time. So here's my favorite detail in all of it. When my agent reads a receipt, you know, and like I said, it it it can't tell, or it isn't sure where the number belongs, you know, it doesn't it doesn't guess. You know, and then it's just it's one question on the Monday briefing, or on the next day saying, hey. This came in.
Speaker 1:I don't know, you know, where it is. So now run all of those against the three questions. You kind of get and understand, you know, how this is, you know, beginning to all hopefully come together. So I keep coming back to this, and I'm gonna say it one final time plainly. I'm giving you a shell and a way of thinking.
Speaker 1:So I've told you how to do it. I've shown you how I do it, My Gmail Plumbing, My Kill List, My Gym, as examples of, you know, that. You can use any of that. If you dread something, if you repeat something, build around those. And six months from now, your system won't look like mine, and that's exactly the way that it's supposed to be.
Speaker 1:So before I land the plane, some straight talk because I'd rather you hear the limits from me than discover them and be annoyed. So this technology is sometimes confidently wrong. So an agent can state a thing that isn't so smoothly, with a straight face. This is exactly why the constitution exists, and it's why my rules, you know, say never guess. You know, it's why sending, email sending stays gated behind my go ahead.
Speaker 1:So it's why, you know, anything that's irreversible, anything touching money, anything touching people gets a human look first on on stuff. So you should you should assume that yours will occasionally be wrong too, and build like that. Trust but verify, and write the verifying rules for it. So second thing is your agent gets better, but not by magic. Okay?
Speaker 1:It gets better because you correct it, and the corrections get written down. Remember one of the starter rules in your constitution says exactly this, that when the owner corrects you, write it down so you never have to do it again. That's called compounding. Early on, you'll correct it weekly. A few months in, you'll notice that you've stopped because the file, you know, reads every morning, contains everything that, you know, that you've ever caught, or everything that you've ever ever taught.
Speaker 1:So and on the leash, start short. Let your agent read before it writes. Let it draft before it sends. Let it propose before it schedules. Lengthen the leash as trust builds, one permission at a time.
Speaker 1:So that's not, you know, being timid. That's, like, a limitation that is a healthy limitation. So that's just how I run mine today, you know, months in on purpose. And the the goal for me is never full autonomy. The goal is leverage with kind of still that human in the loop.
Speaker 1:So let me close with, you know, the question my own system is instructed to ask me. It's in the constitution. The question is efficiency toward what? So, guys, here's the trap, and church people, of all people, should see it. You can use all of this to become incredibly efficient, and then pour every saved hour right back into the machine, and end up the same exhausted person with a fancier system.
Speaker 1:A a faster hamster wheel is still a hamster wheel. So every hour that this gives back to you is only worth something if it goes somewhere. Your family, your calling, your health, the thing that you keep saying that you'll do someday. So people ask me what this has actually given me, and they expect me to say time, but the honest answer is impact. It's not that I have AI agents working for me, so now I get to sit home and watch more TV.
Speaker 1:It's that everything routine, everything an agent can carry, I let it carry. So my hours and my energy go to the things only I can do. The conversations only I can have. The decisions only I can make. It's like having a staff of easily trained hands.
Speaker 1:And here's the best part. If I don't like what one of them is doing, I don't have to fire anybody. I just train them a little better. So one more thing, and it's a small ask. This audio.
Speaker 1:You're welcome to share it freely. K? You can text the link to anybody who'd get something out of it. That's why it exists. But the kit itself, the folder, the files, the app, I'm asking you to not pass those on directly.
Speaker 1:If someone hears this and wants the kit, I'll have them email me, you know, danny@myrevcoach.com, and I will send it to them personally. So you've got the folder, you've got the app, you've got the chief of staff in there waiting and about 10 questions to get you started. So go introduce yourself to your chief of staff. And finally, I'd like to personally thank Tina for helping write this script, producing the audio, and preparing the YouTube post. I couldn't have done it without her.