If you run a business, you say it all day. I'll call him back next week. I'll send that over Friday. Most of those promises live in exactly one place, which is the owner's head, and that works right up until the volume gets high enough that one slips.

We built a client of ours a way to stop carrying them. He types a promise into a private Slack channel the way he would send a text, AI turns it into a tracked item, and every morning he gets a summary of what is overdue, what is due today, and what is still open. He did not have to learn a new app to use it.

Read the full transcript if you would rather not watch.

Jump to a section of the video

This is a volume problem, not a discipline problem

The owner is the one person everything routes through. Customers, vendors, employees, the bank. Every conversation produces a small commitment, and none of them feel worth writing down in the moment, because each one on its own is easy to remember. It is the pile that gets you.

And the dropped follow-up almost always costs more than the follow-up would have taken. The customer who waited a week for a call that never came does not remember that you were busy.

How it works

He types one sentence into his private channel. Something like "told John I'd follow up next week." No form, no fields, no app to open.

The AI reads it and pulls out the pieces: who it involves, what the commitment is, when it is due, and where it came from. That lands in a simple table, which is the part he never has to look at unless he wants to.

He can also ask it things. How much of a product is on hand. What is on the calendar today. It can handle the internal kind of reminder too, like the day two meetings mean a trip needs booking, and it can pull flight details and wait for him to approve or deny them before anything happens.

The morning summary is what makes it stick

Every morning a message shows up: good morning, here is what is overdue, what is due today, and what is still open. Items can carry buttons, so marking something done or opening the related email takes one tap.

That summary is the whole point. A list you have to remember to check is just another thing to remember. A list that shows up in front of you, in the place you already are, gets used.

Why we built it in Slack instead of a new app

Because that is where he already lives all day. We could build something beautiful, and it would not matter if he had to remember to open it. Software people do not use is not worth building.

Slack is not the only right answer, and it is not perfect. For someone else the same thing could run over text messages, over email, or as its own small app where you log in and talk to it. The rule is to put it wherever the person already works, not to pick a tool and ask them to move.

Build the system before you hire the assistant

This is the part I would underline. The usual way to get help is to hire an assistant and hope they figure out the systems once they start, which means paying someone to invent the job while doing it.

Some work needs a person and some does not. Building the part that does not need one first gives you a foundation that runs on its own and can be maintained for a long time. Then when you do hire, their job is follow-through and judgment, the things AI is not good at, rather than figuring out where everything lives.

Don't hire someone into a blank role. Build the system first, then hire the person to run it.

What else can plug into the same channel

Once the channel exists, other things can feed it. It can watch an inbox and flag what needs an answer. It can record meetings and push the notes into a CRM. It can draft replies.

One firm rule on that last one: drafts are drafts. You would never want it sending email on its own. It puts something together, you read it, fix what needs fixing, and hit send yourself. The point is to take work off the owner, not to hand the owner's voice to a machine. That is the same principle behind how much to trust AI in the first place: let it do the part where a wrong answer is cheap, and keep a person on anything that goes out the door.

When this is overkill

If you make three promises a week, you do not need this. You can remember three things.

It starts earning its keep when the clients, tasks and employees compound. My rough guess is somewhere around five to forty employees, but it depends more on the role than the headcount. An owner who is also the main salesperson makes a lot more promises than one who runs operations from behind a desk.

The takeaway

The goal is not more tools. It is getting what is in your head somewhere that reminds you, so you can stop holding it and trust that it will come back when it matters. It is also one of the safer ways to start using AI in a business, because a missed reminder is cheap and nothing leaves the building without you.

If you want to see what else this kind of thing looks like, we build AI into the systems people already use, and if the first question is how your team should use AI at all, that is AI training.

If you have questions about any of this, reach out.

Full transcript

What I said in the video, start to finish. It is unscripted, so it reads like talking. Each timestamp opens that moment on YouTube.

0:00 If you run a business, you probably say this all day long. I'll call him back next week or I'll send that over on Friday or, yeah, I'll hit you up next week. And most of these promises, you know, or verbal commitments, they live in exactly one place in the head of the owner, your head, right? So we built for one of our clients a way to stop keeping all of them up in the head and he didn't have to learn a new app or anything like that. I figured why not show you? So this is what we did. Our client has a massive inventories on the road all the time talking to

0:37 people and whatnot. And they use Slack internally a lot. So what we did is we built a virtual assistant within Slack that you with the help of AI having access to a lot of data. Right. So the thing is what I would like to highlight is that the this type of thing doesn't have to only be in Slack. Right? This is just a use case of us working with the client to build within their environment, right? So promises that slip, right?

1:10 These are common things. Hey, I'll call you back next week or I'll send it over on Friday, like these tasks that you commit. So for him, he's able just to type into this Slack really quick. Hey, remind me about this or can you pull up information on this product? How much of it do we have? And can you give me reminders? Right. So here's how it works. You type it in, right? One sentence in his private Slack channel.

1:37 The AI reads it, pulls out who, what, when, where, why, all that. It's tracked. It lands in a simple table, right? And then every morning a summary shows up, right? A done emoji or whatever. And we can put in buttons of approve this is done or open this in my email account, all that type of stuff. So it's all just in one easy location. And what the AI pulls out, right? So you told John at Acme, for example, I'd follow up next week. So John is the who, the what is the follow-up, the due date, and the source.

2:10 So it notates all of that for him and gives him a, you know, a to-do list or whatever. And this can this could also track like internal things, right? like, oh, you wake up and your reminder is, oh, I've got two meetings today. I need to have my assistant book a trip, or I need to book a trip, or you could have it pull up flight details for a trip, and you could either approve or deny it. Like, there's, there's, the variable, variables are really quite endless here. So the part that for me, that I believe that makes this stick is, like, it shows you, hey, good morning, here's what's overdue, what's due today, and what's still

2:51 open, right? This is a very simple way for people. We could build beautiful things for our clients, but we want to make sure that what we're building will actually be used. And that's why we built this one in Slack because that's where he lives, right? And so these are all just examples, right? But it could get your brain thinking, as a business owner, oh, like, I could have things tailor made in the environments that I usually work in and operate in to do things for me. Like, there could be a foundation to outsource a lot of the things that are stuck in your head as an owner.

3:25 So why we chose to do it in Slack and not a new app, it's because it's a centralized location. Is Slack perfect? No, but like, it could be in a text message format. It could be in an email format. It just depends. Or you could have an actual app developed just for you where you can just log in and talk to the virtual assistant, right? So that's that's kind of, I thought it was a cool use case because it was a very practical. It's like this guy is super busy. He has a lot on his mind. And he just wanted to outsource some of the things that were up there.

4:03 Right. So we build the system before you hire. And what I'm talking about here is like there, there are some things that you, humans need to be involved in. There are some things that they don't. And in this situation, building a system that doesn't need human intervention all the time is better, better to have in place before you bring in an assistant, for example. Right. So it's just a better foundational layer, right?

4:31 So don't, you know, I had to put up here. Don't hire someone into a blank role. Build the system first and then hire the person to run it. Right. So we could essentially build a virtual assistant and a foundation for it to be maintained for a long time. And then anything that the AI isn't great at or the system's not great at, then you outsource the rest to an actual assistant. So what else this can plug into in the same channels. You know, it can, it can monitor your inbox. It could record your meetings and plug into your CRM.

5:08 In essence, the AI, what it drafts, like what it puts together, you would never want to push automatically, let's say for emails, but you could look through it, the draft and then edit it and get it ready to the point that you want and then hit send. It's just supposed to help reduce the workload on the business owner. So when is this overkill? If you're only doing three promises a week, right? Like if you're only like, oh, I only, like, if you're only talking to three people a week, like, you should be able to remember all that. But if you have a compounding amount of clients and

5:44 tasks and employees, this is a great place. You know, I think people who are in that, you know, I would say this type of solution would probably be best for people who have five to I'd say 40 employees maybe it depends on the type of business you have but and then it depends on the type of role that you play if you're in sales and stuff like that but this is definitely something to consider it's not for everyone but yeah it can it can be of high value for certain people so the goal is isn't to have more tools it's it you know it's getting what's in your head somewhere that reminds you and somewhere that you can just safely know that it exists and you can call upon it when you need

6:31 to. So I just wanted to share this whole thing because I thought it was a great example of I know many business owners who are super chaotic and have a lot going on and helping reduce that chaos so they can focus on higher value initiatives was it's it's just a great thing to think about and consider. And so this is a great use case for business owners of a small, small way that AI could really help you run your business in a practical way that doesn't have a potential for large consequences, if that makes sense. So in this scenario, like just having a Slack channel with an AI agent that's able to monitor your emails and maybe tell you some information from

7:18 your inventory or whatever. And so this is just a safe path in my opinion. Anyway, thanks for watching. My name's Tim. I'm from Norsoft. I'm a partner there. and I'm just trying to create content that helps empower business owners to understand what's possible with AI these days. So thanks for watching. Bye-bye.