How to Integrate AI into Your Small Business (Without Breaking the Bank)
Thinking about using AI in your small business, but overwhelmed by the hype (and the price tags)?
In this episode, Dauntless XR co-founders Lori-Lee and Sofia get real about what it takes to successfully implement artificial intelligence as a small, non-technical team. From boosting organic website traffic with AI-powered SEO content to learning where automation didn’t deliver ROI, this conversation is packed with actionable insights, honest lessons, and a roadmap you can actually follow—without blowing your budget.
Episode Highlights
Why Most Small Businesses Aren’t Using AI Yet: Cutting through the noise and addressing real adoption barriers.
Where AI Delivers ROI: How Dauntless XR grew organic traffic and content output using AI for SEO and marketing.
Where AI Didn’t Work: The hard lessons from automating government proposal writing.
How to Build an AI Roadmap: Practical steps for identifying pain points, setting KPIs, and picking the right tools.
Moravec’s Paradox Explained: Why AI excels at data and logic but struggles with emotion and nuance.
Favorite AI Tools for Small Teams: Real examples and honest reviews of what’s working now.
Advice for Getting Started: Free trials, KPIs, and when to walk away from tools that don’t deliver.
Key Takeaways
Start with your biggest pain points and set clear success metrics. We put our thoughts on how we did this on paper so you can steal our process.
Use AI where it saves time on repeatable, rules-based tasks.
Keep a human in the loop for content quality and emotional resonance.
Don’t be afraid to drop tools that aren’t moving the needle.
Revisit your AI stack regularly—what works now may change as tools evolve.
AI Tools Mentioned
ChatGPT (OpenAI): For brainstorming, research, and content drafting.
Descript: AI-powered video editing from transcripts—great for repurposing podcast content.
Notebook LM (Google Gemini 2.0): Secure research assistant for analyzing proprietary docs and finding connections.
Sintra AI: Small business-focused AI agents for SEO, business development, and proactive task management.
Poppy AI: Visual, mind-mapping style interface for content planning and script writing (with integrated GPT-4, Claude, Grok).
Microsoft Copilot: Used at the prototyping stage for software development.
Frequently Asked Questions We get from other Small business owners
Q: Do I need to be technical to use AI in my business?
A: Not at all! Most modern AI tools are designed for non-technical users. The key is knowing your business pain points and being willing to experiment.
Q: What’s the best use of AI for a small business?
A: Content creation for marketing and SEO, data analysis, and report generation are great starting points. Tasks with clear rules and patterns are best suited for automation.
Q: Where should I not use AI?
A: Avoid using AI for highly nuanced tasks like sales calls, complex customer support, or writing high-stakes proposals—especially if human connection or deep expertise is needed.
Q: How do I measure if AI is worth it for my business?
A: Set specific KPIs (like increased organic traffic, more posts, or faster prototyping). Test tools with free trials or monthly subscriptions, and be ready to walk away if you don’t see results.
Q: Is it safe to upload my company’s confidential data to AI tools?
A: Only use tools that guarantee your data won’t be used for model training (e.g., Notebook LM). Always read privacy policies before uploading sensitive information.
Want to go deeper?
Check out another Becoming Dauntless Podcast: Startup Accelerators - are they worth it?
Sign up for the Becoming Dauntless Newsletter to get a recap of news we’re following, updates about the business and more each week https://dauntless.beehiiv.com/
Episode Transcript
Lori-Lee: YouTubers, LinkedIn influencers. The TikTok business gurus have all been screaming at us from our screens to integrate AI agents scale with AI, optimize your workflows with AI.
But if you're a small business or if you are a solopreneur. How do you actually do that? And what does it look like and how do you do it without breaking the bank? 'cause these subscriptions add up so quickly. So today we're gonna break down how we onboarded AI without running up a huge bill. And we'll talk about where it worked, where it didn't, and then kind of the roadmap for how you can do it too.
Sofia: I'm so excited to talk about this today because the number of, at least, friends in my orbit that I talk to that aren't using AI on a daily basis, and I'm like, let's convert you. This is so helpful, but I think there's a couple of assumptions. I don't know if this is true for the people you speak with about this, but you either have to be like an AI startup. Or a tech startup, or you have to be technical, like you have to be a prompt engineer or something to get the benefits of using ai. We're gonna focus on small business applications, right? I am excited about talking about what are our top tools and methods so that we as small business owners get more done in less time.
Lori-Lee: No, exactly. I think the whole point of most of these AI tools is that you are non-technical and able to tap into technical capabilities. Interestingly, when I was doing a little bit of research on this, I did, you know, some very scientific inquiry on Reddit
Sofia: scientific.
Lori-Lee: Yeah, I mean, in a way, in its way.
And there was a thread dedicated to small businesses using AI tools. And, you know, I was going through it expecting to see like the just, you know, I. Every random AI agent, GPT wrapper tool you'd ever heard of. Surprisingly, the most common answer was, I'm using chat GPT to ask questions. So I think a lot of people are talking about this big AI revolution, small businesses like, you know, giving, new companies and new startups and creators of Bump, but I don't think there's a lot of people actually doing it. Just based on the admittedly limited research I've done. It doesn't seem like it is as prolific as people make it out to be. Even if they're evangelizing AI and saying it's the best thing ever.
I don't think it's quite as widespread yet I guess the best place to start would be cutting straight to the good stuff. Where have we seen an actual ROI from using AI in our business? Keeping in mind, like we run a software tech company.
Sofia: The number one area where it's been really useful for us and we have the metrics to demonstrate that has been marketing planning and specifically SEO content generation, our main goal with marketing content and SEO content generation is organic traffic growth, right? We wanna see organic traffic to our pages, to our posts on social and to our website go up. When it comes to creating this content, we go to the resources that people already know about, right? So Google Trends, answer the public, things like that, and then we feed all of that content to an AI agent. I'll talk later about which one we like for this, but you can do this with most AI agents.
And then based on the context that an AI agent has been provided and out of other context that you give it about your business, we ask for things like a marketing plan. We ask for things like ideas for long form. SEO content that should go on our website, we ask what, based on all this information we've given you, what are the keywords that you recommend us doing?
You can even feed it your own website and say, what keywords are you already finding on our website, and what keywords are you not finding on our website based on this other material that I'm giving you? So it's, for us, it's like having a research assistant that knows exactly what's trending, but for only the content that is relevant for us.
Once we find those ideas, we will usually take one of the prompts that it gives us and then write the outline ourselves. I think this part is really critical. The outline is the backbone of the content, and I think it gives the AI direction and keeps the content focused and valuable.
And then we give it to the AI agent, have it feed us a draft, and then edit that draft and post it. For I would say Q1 of 2025, if we are comparing our organic traffic performance from. from. Q4, or Q1 of 2024. We've had significant growth in the traffic that we're getting.
So we are pleased with the use of AI for marketing content for us.
Lori-Lee: So we've seen positive ROI as far as market research helping us produce more content, which then leads to more eyeballs on our. Brand on our product that we can convert into, you know, whatever we want. Sales, email subscribers, leads all that good stuff. Which is interesting because that's not necessarily even the best applications of ai, but we'll get to that later.
Sofia: We've had this discussion a lot, right? Like what humans are really good at is storytelling, eliciting emotions. And content is a way that we do that. So I think the reason why it works for us is because we feed it content relevant to our brand and the topics that we wanna speak to, and we write the outline ourselves.
I think if it were a hundred percent AI generated, we might not be achieving the level of ROI that we're getting, but
From your perspective, where did we try to embed AI into our small business operations and it did not deliver an ROI.
Lori-Lee: Yeah, it's a good question. And there's one, blaring answer, and that was in government proposal writing. So we tried to automate as much as we could, our government. Proposal writing for contracts with a dedicated AI product. So this was not like we tried to use chat GPT to write, you know, a crazy amount of proposals.
All that have different requirements and different outlines and things like that. We had a dedicated proposal writing software. It did work in that we were able to produce what I thought was a good volume of quality. Proposals, we put our own final edit on it so it wasn't, you know, just a hundred percent AI generated, definitely allowed us to increase our capacity, but we won zero of those contracts and I.
Sofia: Yeah, it was bad
Lori-Lee: Yeah, it allowed us to, at least double our proposal output. But then we won none of the contracts and we went back to winning contracts when we stopped using the proposal writers. So there was a very clear cause and effect there. Or root cause analysis of why our win rate suddenly tanked.
And I really don't understand why, because I'm not gonna let anything out the door written by me or written by ai that's not to our standards. I thought they were too standard, and I don't know that someone who was looking at it from the outside could tell, oh, this one was AI generated versus this one wasn't because we edited them before.
They're submitted, And we outlined them.
Sofia: That we talked about in marketing content.
Lori-Lee: Mm-hmm.
Sofia: We made sure that we were providing relevant reference material that we were, providing prior winning proposals as reference material. Like it felt like we had done it responsibly.
Lori-Lee: Going back to where we have seen ROI do, you know, like from the development side of things, 'cause we also write software, so we have software engineers writing code. Do you think that has caused us to be able to work faster or better? We haven't actually seen an ROI yet, but
Sofia: haven't seen an ROI yet because the products that we have out, were not built using any AI coding tools.
Lori-Lee: mm-hmm.
Sofia: We are working on something right now where we are using Microsoft Copilot. It is at a prototyping stage, which I think is the right stage to be using an AI agent for development.
Lori-Lee: Mm-hmm.
Sofia: And so far the feedback from the dev team is that it's useful and it is helping them. Prototype faster as we're trying to test out different things, so I think we'll see whether or not the ROI comes, but I will say we are trying it, right? Somewhere where it's working for us, it's definitely on the marketing, specifically on SEO content to drive organic traffic, not working government contracts that that coffin has been closed.
Lori-Lee: Yeah.
Sofia: We will not let an agent touch that stuff again. And TBD would be development.
Lori-Lee: Yeah. Okay. No, I just thought of that because I'm sure that's one of the things a lot of people want to use AI for is coding assistance or just straight out, trying to replace a developer with an AI agent, which I don't think we're there yet. So that's why I wanted to bring that up. But
we won't be letting an AI outright generate an entire proposal again anytime soon.
Sofia: Slightly more positive note. What are our favorite AI tools right now?
Lori-Lee: You know, going beyond chat, AI that is really enjoying descript. We're using it to record this podcast right now, but it is a video editing tool, video editing software that allows you to edit from a transcript.
So you upload your video, turn it into a transcript, and you can, you know, if you delete a word, it will delete. That corresponding piece of video. It has a feature in it called Under Lurd, which is like AI assistant and IT underlord can do all sorts of things, remove filler words, shorten word gaps.
You can turn on the green screen, turn it off. It also has a repurposing tool there. And this is not any, any shade on script. I think all of the repurposing tools. No matter how well they're marketed, they're just not that good yet. Like AI is not good at pulling something salient or conversational or whatever out of a piece of video because it doesn't know what that is.
So it's really just hit or miss when you hit that button if you're gonna get something usable or not. But anyway, I love editing. Using this software. It's allowed me to like 10 x video output. It's so much faster. Love it. One I haven't tried yet, but I think I will love it. I just heard about it this morning, so like, if it turns out it sucks, like don't come for me, but Poppy ai and I think I'll like it because it has a different interface.
It looks like a Miro board. If you've ever used that software, if you watch our podcast episode on nasa. We have a Miro board up, but it is more of a mind map style interface where you can drag and drop in different resources and then input your prompts and have it generate like it's more pre-production stuff.
So like scripts, video titles and thumbnails or outlines or I'm sure if you wanted to, you could write a whole ebook in there and it has. Chat, GPT, Claude and Grok, all the paid subscriptions all rolled into it under one. So if you get Poppy, you can cancel your other ones, which I also really like. And it's a bootstrapped company, which I'm a big advocate for, but I haven't actually tried it yet.
I've just been doing a bunch of research, but I have a feeling I'll really like it.
Sofia: Oh, I wanna try that too.
The idea of being able to lay it out visually. I have two favorite tools. I don't think I can choose between them. So I'm gonna talk about both. The first one I'm gonna talk about is Notebook. Lmuses Gemini, I think. 2.0 and it is a research assistant. I like it if you provide the notebook with your sources. So your website PDFs you're using, and it can help you find relationships between those documents to answer your questions and help you create content. Pros to notebook. Is the data that you provide, it is not used to train the main model. So you can provide it like proprietary company information, proprietary information that you're trying to work with that you couldn't provide, like a free tier chatGPT, right?
They're not the only ones that could do this, but it is a pro. The other thing I like about it is. That if you prompt it with something that is not in there, like in not something you provided in your notebook, it'll just tell you like, there's nothing in the notebook that I can refer to, so it won't hallucinate.
It'll just say, give me better resources or gimme more resources. So I love that about it. The other favorite I have right now is Sintra ai. So they use open AI's, GPT-4, I'm pretty sure, essentially it's as if they took ChatGPT and then trained it on addition. Small business specific content and then gave you access to that agent. So if you took GPT 4.o and trained it on SEO. You have an SEO agent, you trained it on biz dev, you have a biz agent and it, they brand it as like your employees that never sleep, which is such a great way to do it. What I like about Sintra specifically is it is the only AI tool that I'm using right now that is proactive. In interacting with me. So when you log into Sintra in the morning, it asks you more context questions. It's always trying to build more context about your brand and your business and what your goals are. And then b, it will prompt you with, it's been a while since your SEO agent has done an audit of your website. Would you let it get to run an SEO audit of your website and provide you with a report? Hey, the new month is coming up. Would you like us to generate a social media marketing plan for you? And then it, you can just say, yes, and it will go do it,
Lori-Lee: And it will also tell you if you're using the wrong agent. I dunno if you've made it, do that. I do that all the time.
Sofia: I haven't done that.
Lori-Lee: Okay. Well, it yells at me occasionally and if I'm talking to the copywriting agent, which is called Pen, and the questions I'm asking start swaying more into SEO, it will tell me, you would be better off asking Seomi who's the AI agent or if I just click on the wrong one and start typing.
Something not relevant to what that agent is trained on. It just will redirect you.
Sofia: I do wish you could add multiple agents to a group chat and be like, Penn, I want you to copy edit this blog post I wrote and then Seomi can you audit it for our target list of SEO keywords and tell me if there's any that I'm missing or that I should add. Like that
Lori-Lee: Yeah.
Sofia: So nice. Maybe in the next iteration we'll see.
Sofia: If we were starting out over again in terms of bringing AI into our business. What do you think an ideal AI implementation roadmap would look like?
I've been thinking about it a lot. I've been asked about it a lot, and I actually have created a roadmap document that we can share. I think the secret to a successful AI implementation is actually a concept that's been around since 1983. And if you can understand this, it makes trying out AI so much easier.
And I'm talking about Moravec's Paradox. Moravec's Paradox is the assumption that things that. We humans find it easy like walking or goal setting or paying attention to something interesting.
We assume because they're easy for us, that they require very little computational power for an AI to do. If a one or 2-year-old human can do all these things, why can't an AI with unlimited processing power do them? We also assume that. Things that we find difficult like playing chess, which is at least something I find difficult or like recalling large volumes of data or solving a complex equation.
We assume that because they're hard for us to do that, they require a lot of computational power. The paradox is that the opposite is true. AI completely tears through mathematics, engineering games, logic, scientific reasoning, but it is not good at anything to do with perception, attention, visualization, tone, sarcasm, motor skills, social skills, emotional, anything, emotional intelligence. Eliciting emotions, any of that. So if you think about it this way, AI can tell you the exact dimensions of a room and how to reverse engineer that room, but it cannot read the room.
And it's not that it never will, it's just if you're rolling out AI tools right now, this is the reality that we're dealing with. So I think the first thing you have to do is keep in mind Moravec's paradox when you're figuring it out. What your AI rollout plan is gonna look like. The first step to that is to figure out where the biggest pain points are in your business.
For us it was like creating content at scale. We needed volume of content. It might be maybe the biggest pain point is, you know, your sales funnel needs optimizing, or it's marketing or, it's developing a new. Website or app or something like that. So you figure out first where there's room for improvement within your business and where you're feeling it the most.
And then from there, set your KPIs or set your metrics that matter. So you have to decide before you go and experiment with all these AI tools, how are you gonna measure if they are making an impact or not. So like in our case, I know I keep going back to our content example, but.
We just wanted to be able to post more. If we were successfully getting more posts out there, that was a win for us. Right. And the example of code, like if we're able to get to prototype faster. Than what we would normally. That's a win, right? Set your KPIs and then map your pain points to the ones that are a good fit for ai. Going back to. Moravec's paradox. Poor candidates for AI automation are things like customer service tickets, product management, company strategy, original messaging, doing sales calls, which is all hilarious because when I was researching for this, I found a HubSpot blog post on implementing AI in your business.
And the top two things it said to use AI for were customer service and sales. And I was like that. That is, I'm sorry, HubSpot. That is terrible advice.
As someone who worked in customer service, you have to split help desk type transactional support like true customer service. So like, yes, maybe there are, you know, some things where an AI chat bot can help, but the vast majority of them probably not.
Especially if you're a really small business. I don't know that you really wanna put too many layers between you and your customer too quickly.
Lori-Lee: No, I mean, you wanna be talking to them. I had to reach out to customer support for Gusto, which handles our payroll the other week, it makes you type in, like, what is this about? And it was about, you know.
Handling payroll, handling taxes, making payments to government agencies, like important stuff. I'm not writing to complain, like I need to reset my email password, or my login password. And it put me on a chat with an AI agent and I wanted to throw my computer out the window.
It was. Terrible at answering my questions and they tried to hide that this was a chat bot too. They tried to make it seem like it was a human. And I'm like, this is very clearly an ai. You're not fooling anyone. And it would not put me in contact with a real human. And I still have not got that problem resolved.
And it's really annoying. So do what you will with that information, but I certainly will not be using AI for customer service or sales. My business, let alone any of those other strategies.
That's poor candidates for AI automation. I think good candidates for it are things like doing anything that ends in calculation or has analysis in it, so like cash flow calculations and reporting. In general report generation, if you're able to feed it a bunch of data sources and ask it to generate a report, connecting heterogeneous data sets, doing rough cuts of video content.
And the thing that all of these have in common is they're all tasks with very set rules. Data processing is going to have a set formula that it follows pattern matching, follows a set formula. So those are the kind of tasks that you could completely automate with AI that I think are really good places to start.
We didn't really have too much of this in our business that was eligible. For AI automation, we could do a little bit probably with sales funnel analysis, having it look at our sales funnel and tell us maybe not what to do, but tell us where things can be improved. And then I have the kind of like, meh, so, so candidates for AI automation, and in my opinion, that's code generation or coding, support, content production, and content repurposing.
We still use it for all of those things, but there's definitely a human in the loop. It is not to the point where I just hit boop, repurpose and it spits out five videos that are fantastic. Like, absolutely not. But it does shorten the time it takes to do all of those things. So once you have mapped what out of these buckets you are going to try to implement AI into, the next thing I would say is go try the tools that are free, or do a one month subscription if a one month subscription is available. And for most of the ones where it isn't, they at least give you 30 days to cancel if you don't like it. But try the basic stuff first. And then if you see an improvement based on your KPIs, you can.
Go a step further and get the pro tier or the subscription or whatever that is. But that's why it's important to have those KPIs up front. And then, when you have your winner, you can commit to it and know that it's going to be an investment that will pay off.
Sofia: Once you've committed, a method to reevaluate. Rate, because some of the KPIs you're not going to hit within 30 days.
Lori-Lee: True.
Sofia: would be from the content generation, right? Organic traffic is lagging. I feel like we had to do it for a while before we could really evaluate that, you know, partially or mostly AI generated article that we posted to the website.
Is it helping? Right? And with your example of the proposals. It took 90 days, you know, four to six months before we were realizing it. But because we had that KPI in place of that, we're only willing to use AI to bid more if we can maintain our win rate. It was easy to walk away from that because I. The success metric wasn't there. Right. I was, you know, as much as our, I'm always proud of our successes. I'm proud of our, the way we handle failure too, and the way that we just looked at that and we were like, not gonna sunk cost fallacy. Keep trying to make this work. It didn't work. We're a small business.
We don't have that much time to waste. We're gonna walk away and try something else.
Lori-Lee: Yeah, exactly. That's probably some supporting material right there to maybe prioritize products that either have the free trial or that you can cancel month to month. So if you find out like, this isn't moving the needle, you can just get rid of it. I know most of them offer that, so, you know, you don't have to be too scared about making that initial, add to cart purchase.
Sofia: It will be interesting. What I would like to do is come back in six months, in maybe a year see how we're using AI and what our favorite tools are and where it didn't work for us in the business. Because I think if I had listened to. conversations we were having now six months ago, I'd be like, what? We're using it there. It didn't work out for proposal writing. Like, I'd be very surprised. Right? So I think we should revisit this at least by the end of 2025 and be like, okay, where is it working? Where is it not working? From the perspective of one small B2B SaaS company.
Lori-Lee: Yeah, absolutely. I think we'll be surprised and now we have documented evidence.
Sofia: Of what we thought at the time.
Lori-Lee: If you enjoyed this episode, be sure to check out episode three where we break down winning our first NASA contract with no experience and how you can do it too. And by the way, it was a proposal that AI did not write so we'll see you next time.
