You want to build AI tools for your business. Here's what you build first.

Almost everyone I talk to says the same thing. "I want to build something with AI but I just don't know what yet."

Most people in this position do one of three things:

1. Take endless workshops on how to use AI, but still never get beyond implementing the exact thing they learned in the workshop
2. Build tools based on where the dopamine is, like morning briefings or dashboards or fun little shower thoughts ideas, or
3. Stay completely stuck and build nothing, or start feeling like it's just not the right time instead of just not having clarity about the right thing.

What do I say when someone asks me this question?

"Where do you and your clients get stuck?"

The friction points in your business that keep you from doing your best work and stop your clients from implementing it? That's where you start. And trust me, there are always friction points.

I'm an Anthropic Certified expert who launched a successful micro-SaaS and now help other founders build their own tools. Here's what I've learned:

12 truths about building tools and tech from a founder who learned the hard way.

1. Chatbots turn your knowledge into a commodity. I mean, AI did this already. The value of information is near zero. If an AI chatbot can give you the information, point to sources, and tutor you through it, people will not buy your courses or workshops doing the exact same thing.
   
   Turning around and making a custom chat interface that delivers your knowledge doesn't fix the problem, it tells people that what you do can be outsourced to a chatbot. It devalues the other parts of your work in the process.
   
2. It completely misses the value of tacit knowledge, lived experience, instincts, rapport, and trust. Researchers at MIT Sloan identified five "uniquely human capabilities that AI cannot recreate."(Source paper here.) They are:
   
   Empathy and Emotional Intelligence
   Presence, Networking, and Connectedness
   Opinion, Judgment, and Ethics
   Creativity and Imagination
   Hope, Vision, and Leadership.
   
   Your services can't be replaced by a chatbot unless your services didn't require any uniquely human capabilities in the first place - in which case you have bigger problems in the days of near-infinite access to information.
   
3. People will always value done-for-you work more than do-it-yourself, unless it's for enjoyment. Don't get me wrong. I love knitting, weaving, building LEGO, reading books, sewing, 3D printing... I'm not going to outsource those things just because I can. I enjoy the process.
   
   When it comes to things I need done in my business, however, I will always value someone doing it for me more than learning to do it myself unless it's something I genuinely enjoy or need to know to deliver effective services.
   
   People don't want your Facebook Ads course anymore unless they're trying to become Facebook ad managers. They'd rather have a Facebook ad campaign setup wizard that applies your framework for them.
   
   And that's probably not a chatbot.
   
4. Automation just grew wings. For almost a decade now, every time I'm working with a founder and they are overwhelmed, I ask them three questions about every one of their tasks:
   
   Does this need to be done at all?
   Does this need to be done by a person?
   Does that person need to be you?
   
   If it doesn't need to be done, we eliminate it. If it needs to be done, but not by a person, we automate it. If it needs a person but not you, personally? We delegate it. The rest is your highest and best work, and where you should focus for the greatest impact. Simple enough.
   
   That middle section though? It just got a LOT larger. A lot of things we used to need people for, simply aren't human-required tasks anymore. Things like data entry and repetitive manual tasks that were complex and difficult to automate a couple of years ago are now an hour building a custom script, and done.
   
   However.
   
   This means that a lot of people are automating things without asking the first and third questions. They're automating things that don't need to be done at all, and things that need to be done by themselves personally. Things like human connection or answering questions or knowledge delivery.
   
   Often through things like chatbots.
   
5. Chat interfaces have specific things they're good at. If your clients or customers have complex decision making or information gathering to do, and multiple variables at play? Chatbots can help. If you need to send similar messages to lots of people but there are a bunch of variables in exactly which messages go to which person? Chatbots can help.
   
   Most other things, chat interfaces actually get in the way and cause increased frustration.
   
6. The shape of your tools should match the task. If you have an identical process that repeats itself, you need a checklist. If you have a variable process or a setup process that only runs once, you need a wizard. If you need to present information for decision making, you need a dashboard. If you need to collect information, you need a form.
   
   All of these are longstanding conventions in software development that are being replaced by chat interfaces in places that really shouldn't be chat interfaces.
   
7. The first things you should build are much more boring than you think. Putting a custom wrapper on OpenAI or Anthropic products and then adding a knowledgebase isn't a tool that actually solves a problem for you or your clients, unless that problem was your clients saying "I don't like paying for consulting time".
   
   But there are AI tools you should be building first. They're just a lot more boring.
   
   Second brain systems like Notion or Obsidian, client onboarding processes or dashboards, project trackers, little scripts to automate $10 tasks... These are the early builds most people should create because they solve actual problems.
   
   Your work will both accelerate and improve when you create them, and the customer experience will get better, too.
   
8. Your first build should probably be a second brain. Not just one for you, but one for your AI system, too.
   
   When OpenAI partnered with the US government in a morally reprehensible back-room deal, millions of users switched from ChatGPT to Claude - owned by Anthropic. But there was a big problem.
   
   ChatGPT held all of the knowledge about the user, and didn't give it up easily. In fact, the more people left OpenAI, the harder ChatGPT fought to hold their data hostage.
   
   A second brain for your AI-supported work gives you digital sovereignty. You can move to a new model tomorrow and take all of your context and work with you, without having to fight the system itself.
   
   Obsidian and Notion are two great options for this, especially if you keep the AI-generated content siloed and separate from the human-generated content.
   
9. Don't build toys for dopamine. I like playing with tech as much as the next person, and as someone with what I lovingly refer to as raging ADHD, I am much more likely to play with a new toy and build something that gives me energy than I am to work on the boring day to day bits.
   
   However.
   
   Those boring bits like client dashboards or project trackers or second brains or intake forms or interactive lead magnets? They're going to make the rest of your work more efficient and more effective pretty much immediately.
   
   This both improves your business and frees up time to play with building fun things for the dopamine, too!
   
10. A tool in someone's daily workflow is the best retention mechanism you've ever had. Think about it. If you sell access to your time or your brain, that's easy to cut if things get lean and hard decisions have to be made.
    
    If you sell actual implementation that a client relies on, that's harder to cut as long as they still need the work done. Sometimes they'll find a way to bring in house or reduce the workload, but it's better.
    
    But the ultimate client retention? Owning a tool that they use in their personal workflow every single day.
    
    If you own the CRM or the task management app or the social media scheduler that they run every day of their lives, it's a lot harder to walk away from that. They have to not only find the knowledge (your brain) elsewhere, but also the implementation (the time) and the tool to get it done. And then they have to actually migrate the data!
    
    If you've ever switched website platforms, you know this. Not many people go from Wix to Squarespace without just straight up rebuilding everything from scratch, and there's a reason for that.
    
    Make your tools indispensable.
    
11. You don't need to know how to code, but you do need to understand the basics of privacy, security, and structure. Most of the terrible vibe-coding apps that exist today aren't bad ideas, they're poorly executed ones.
    
    People who don't understand privacy and security building authentication that can have user data stolen. New vibe coders exposing their API keys right in their code so anyone can run up their usage and cost them thousands of dollars.
    
    And yes, founders who don't understand hosting, posting links to localhost for their apps and wondering why they get laughed off the internet.
    
    Having a basic understanding of privacy, security, different programming languages and what they do, the architecture of an app, being able to "read" the code produced by vibe coding tools will put you light years ahead of everyone else with just a few small pieces of understanding.
    
    And they'll reduce the risk of getting ratioed on Reddit or Threads by 10,000%.
    
12. Stop trying to build everything from scratch. You don't need to build a clone of XYZ thing. I promise. Someone already did, and there's most likely an open source version of it or at the very least a library you can install.
    
    Stop scratch-building websites, and start getting AI to build custom themes for your favourite CMS like WordPress or Joomla or Ghost.
    
    Stop scratch-building marketing and mailing list tools and go install Mautic.
    
    Quit trying to find a replacement for Skool and install Discourse.
    
    End your build of a Thinkific replacement and go install Moodle.
    
    There are open source alternatives that allow commercial use and modification all over the place. Check the licenses to make sure you can modify the code and charge money for accessing your services, and then use them! Contribute code to them. Donate and volunteer.
    
    The vast majority of open source projects are run by non-profits and volunteers who donate their IP, time, and energy to the common good.
    
    You don't need to build it. You just need to know who already did.

Right now, I'm bringing together a group of founders who want to grow their consulting or creative business by creating tiny tech tools to automate their marketing, improve their client delivery, and get better results for their clients, without losing the high-touch expertise that makes you valuable.

We call it human-first, tech-supported business, and we're building it together in Solo School.