How to actually use AI in your business

Jorge: The cold open was “I don’t think humans were created or meant to be on computers.” Tell me more about that.

Nicolas: I’ve had this thought before. I don’t think we were built to be in a 9-to-5 looking at a computer all day, 50 weeks of the year. But the main advantage we can get from AI disruption is that it frees us up to focus on what actually matters — human relationships. The more relationships you can build, the better you do. Now that we have tools to automate the tedious parts, we can focus on the things that are genuinely human.

Jorge: Tell me how you first got interested in AI.

Nicolas: When AI first came out, ChatGPT was the only real model people had access to — I think 2022, 2023. And I was against it because I didn’t trust it. I didn’t understand it, and I was very hesitant about what it could potentially do. So I didn’t use it for the longest time. Then maybe six or seven months ago, I finally said, okay, maybe I should start looking into it. And the more I used it, the more I realized there was a lot here. I’d love to focus on the things I enjoy doing and try to be 10 times better at them. I’d prefer to let the machine handle what the machine is good at.

Jorge: You said something I want to flag: “don’t rely on AI too much.”

Nicolas: Yeah. I wasted some time over-analyzing with AI — trying to see every angle where things can go wrong, when I could have just focused on productivity tasks and been more intentional and more human. AI is a robot. It doesn’t have that flexibility and that sensibility. It’s a tool to support you, not to make decisions for you.

6:08 Skip to “What happens when clients trust the first response”

Jorge: Can you tell me about the last time you watched a client hand AI a bad instruction and pay for it?

Nicolas: It’s almost every time. I ask clients what they use AI for right now, and they say “ChatGPT, for my marketing, social media posts, writing emails.” And I find a lot of the time when they give it what they need, it gives them bad outputs — and they keep going back and forth until they get something close to what they wanted. A lot of time is lost that could be reduced by understanding how to actually talk to the AI. And some clients will come back and do exactly what the AI told them to do — and the AI told them the wrong thing. People tend to rely on the first response AI gives them. That’s a mistake a lot of people make. You say “Hey AI, I need this and this and this,” ChatGPT answers, you say thanks and go with that. You shouldn’t be doing that.

Nicolas: Ideally, you’re not going to find a perfect answer after your first prompt. An AI is technically not trained to know anything about you, your business, or nothing at all. It’s only trained to take the information it receives. It’s a large language model — it takes whatever info you give it, does its own processing, forms its outputs based on that, and gives you a response.

Nicolas: AI can also make mistakes. The other day I tested it — I asked it a silly question and I disagreed with its answer, which I should have. But then I said “No, you’re wrong and I’m right” — even though I was wrong — and it agreed with me. So AI is not smart in that way. It’s smart, but it will agree with you if you push hard enough. That’s where people need to be more careful.

9:08 Skip to “The 10/80/10 rule”

Jorge: I read somewhere that the first 10 or 20% is you prompting, then it gets the whole job done, and the last 10% is yours.

Nicolas: That’s exactly it. 10% of it is your idea coming to fruition by talking to the AI. 80% it does — it gets it done. The last 10% is you reviewing and making sure that everything is good.

10:30 Skip to “The best use case for a Moncton business right now”

Jorge: Let’s say a business owner in Moncton right now is buried in work. What’s the best AI use case you’re seeing?

Nicolas: Take a property management company as an example. They deal with tenants, leasing, maintenance, new inquiries, calls — all kinds of things. A lot of the time their agents are on the go, dealing directly with people. And in the background, all the paperwork is sitting there waiting. That’s the kind of stuff AI can basically handle for you. The intakes, the forms, the follow-up emails, the reminders — administrative tasks that typically get done by a human, a lot of it can now get done with AI. And we’re seeing a lot more of that now, especially with the newer tools. Claude Co-Work is a good example — Manis is another.

Jorge: Can you explain what Claude Co-Work is?

Nicolas: Claude Co-Work is an automation tool that Claude built. If you’ve heard of Zapier or Make.com, those are the common ones. Well, Claude Co-Work is its own version — you go into the Claude app itself, tell it what you need it to do from your instructions, and it just goes and does it for you. It automates tasks versus you talking to it and then going to do the work yourself. I connected it to all my tools on my computer — all my software, my Mac, my files and folders. When I got a new computer and my files were migrated over as a mess, I went on Claude Co-Work and said: “Go into my folders, look at all my files, and organize everything for me based on these criteria.” Five minutes. Done.

14:19 Skip to “The privacy question”

Jorge: Do you think it’s a valid concern that people have about privacy when talking to these AI models?

Nicolas: I think it is a valid concern. It depends on which AI we’re talking about. When we think about big companies like Google with Gemini, or Anthropic with Claude — they claim they have safeguards in place so that your privacy is secure, and I believe that largely. But there was also a concern for privacy way before AI was around. Google’s been around for a long time. Social media’s been around since the early 2000s. People have been posting about their life, their pictures. Your privacy is already out the window. AI coming into play isn’t going to make that much of a difference. There are privacy protections in place that are no different than every other tool or software you’ve ever used on the internet.

16:05 Skip to “Where AI access is heading”

Nicolas: I’ll be honest — there’s AI technology available right now that’s not available to the public because it’s so advanced they don’t want to release it too fast for security reasons. And I think AI is eventually going to start becoming harder to get access to. The more data centers, the more they release, the heavier loaded it gets. The cost of consumption is going to skyrocket. The companies are going to have a harder and harder time keeping up. Free tiers aren’t going to do too much, and people are either going to have to accept that they need to pay for a decent plan or miss out. I could be wrong. But with the way things are going, I could see AI becoming a professional tool — kind of a luxury. You have to pay the premium price.

19:37 Skip to “Which industries AI will disrupt in New Brunswick”

Jorge: What are some of the industries that can be highly disrupted by AI?

Nicolas: AI technically has a role to play in most, if not all, industries. You could have AI operate in a medical office. You could have AI help diagnose a car problem for a mechanic. The disruption in the New Brunswick market especially would be an increase in job losses — some owners might think it’s more efficient to just have AI do some of the work. But I wouldn’t expect it anytime soon. I’ve even read about business owners that are regretting letting go of employees for AI, because they’re realizing the AI is actually costing them more in consumption and usage. New Brunswick, Atlantic Canada — like everything else, we’re behind. In bigger cities, in the States or even in the bigger cities in Canada, most businesses are already incorporating advanced AI features. We’re just not really there yet.

26:19 Skip to “Why soft skills future-proof you”

Jorge: Let’s say I have an MBA, or I’ve been working in accounting or marketing for years. What should I be doing right now to future-proof myself?

Nicolas: Soft skills, 100%. I really don’t even think people should be worried that AI is going to come for their job. If you have the capacity, the experience, and you’re working directly in the field, chances are people are always going to need a human. AI still needs to be maintained. AI should almost be treated as your administrative assistant, not a full replacement for the current work you actually do. And when you look at what AI outputs — I’ve seen people using AI to design websites, marketing strategies — they all look the same. At first people were excited and scared. But then people started to realize what AI was outputting was not unique. It all looked and felt the same. And then people started to draw back and realize: we still need a human touch on this. AI does not have emotion. It can’t create a human connection.

31:47 Skip to “How to actually prompt: objective first”

Jorge: What’s one thing people can do to improve their prompting?

Nicolas: I follow Dan Martell, and what he says about prompting really helps. When it comes to prompting, you start off by giving the AI your actual objective. So say you want to design a document for a client that describes their marketing strategy. You go on Claude and you say: “Hey Claude, I have a client I’ve been working with. I need to have a marketing strategy on one page that talks about these specific discussion points. That’s my goal. Here is what I need you to consider when you’re creating this document.” And before you even send it off, it’s always a good idea to say: “Please ask me any clarifying questions you may need before building this document. If you don’t have any, go ahead and do it that way.” You start by giving it the end goal. You explain how you want to get there. You let it ask clarifying questions. Then it goes in and does it.

38:04 Skip to “Which AI tools to use and when”

Jorge: People mostly stick with ChatGPT. What else is out there?

Nicolas: That’s something people tend to forget — it’s not just one. You’ve got Gemini. You’ve got Claude. You’ve got Grok. Gemini is great because it generates the best images and it has access to YouTube and all the Google stuff, so it makes things way easier. Claude is the best tool for writing, but also just for all-around work — it’s my main one. I still use Gemini for quick questions, or when I need a recipe for supper. Use different AIs, different models. Don’t be shy to use at least a couple. The Claude Max plan is around $140–160 a month. For most people, Claude Pro is good enough. And they all have free tiers.

42:47 Skip to “If you’re starting a business today”

Jorge: Let’s say someone is starting a business right now. Where do they start with AI?

Nicolas: If they’re starting from scratch, whatever business, I would just get Claude — get a subscription, even just Claude Pro to start. You’ve got your chatbot, your Co-Work, your Claude Code all built in. I would use Claude to help me come up with a business strategy and a marketing plan based on whatever business I want to start. And then from there, after the plan is put in place, I’d say: “Okay, now let’s build a calendar or a schedule for the next 30 days that’s going to allow me to keep on track with this plan.” Because it’s important to have a plan and a schedule to follow — whether people want to admit that or not. Claude does really well at planning with you. But you don’t let Claude do the planning for you. You’re working with Claude, with the ideas coming from you, while having Claude be the backbone or your second opinion.

Jorge: Can you introduce yourself and explain how you help people?

Nicolas: My name is Nicolas Michaud. I own and operate a business called Technica I Solutions. The business is designed to help business owners save time, money, and stress. When I come into a business, I’m not coming to sell you AI. I am here to sell you solutions to your problems — whatever those problems are. If you’re telling me, “What can AI do to help me?” Let’s have a conversation for about an hour, and I’ll tell you exactly how AI can help you. And then I’m not only going to explain — I’m going to do it for you as well. The gap I see in businesses is that they go to presentations and seminars about AI. It’s great. But they still have to go and do the work themselves after. I’m going to do the work for you, do the plan, build the solutions, plug it into your business, train you on it, give you a playbook, and make sure you’re set to go.