You can’t scroll LinkedIn anymore without noticing that every other post is focused on either the horrors or opportunities of AI tools, particularly generative AI. Big business is betting big on the technology, investing billions of dollars and laying off thousands of employees already. For the most part, layoffs are to help cover the bets on AI rather than a reflection that AI is already filling roles directly.
The general feeling pushed in the entrepreneurship spaces is adopt the emerging technology, or get left behind. In fact, incorporating an AI use-case is key to attracting investors in 2025 (with the caveat that the rest of the business fundamentals still need to be strong, AI isn’t a magic bullet). AI startups earned 53% of all global venture capital (VC) in the first half of 2025, and 64% of US-based VC dollars.
Adopt AI as a core pillar of your business or miss out on funding and the future. This seems open-and-shut.
Well, Not so Fast
It’s not all sunshine and rainbows, unfortunately. In the current moment, in some industries, using AI in certain ways can earn backlash from customers. And complete dependence on AI, or AI-induced laziness can result in disaster for your business. Studies from MIT and Microsoft even show that dependence on generative AI can have a negative impact on cognitive engagement and activations in the brain during certain tasks. There’s also the hallucinations (inaccurate or made-up information), which are an ever-present issue in commercial products like ChatGPT and the like.
There’s also still uncertainty about legal ownership of AI-generated content, as well as the legality of using copyrighted works to train many generative AI models. That doesn’t get into the environmental impact and a variety of moral questions regarding how the models are trained, and unintended bias that might get included in results.
If you’re going to adopt AI in your business, you really do have to weigh all of this before you decide if it’s the right move for you. Some of these issues are very real, and some are mitigated or even eliminated depending on what specific AI technology and use-case processes you’re implementing.
On AI as a Shortcut
“I learned in the military that shortcuts get you hurt. Vibe coding is a shortcut, and it encourages your team to skip the discipline required to build something that lasts. What you’re left with is a black box running a critical part of your company. When it breaks, and it will break at the worst possible time, you’re left facing a crisis with a team that isn’t trained to handle it. You can’t build a serious operation on a foundation nobody understands. Real capability is earned. You can’t just prompt for it.”
–James Suh, Marine Corps & US Navy Veteran
Founder & CEO of Nashville Analytics
What Even IS This AI Stuff, Really?
When we talk about AI, it’s an umbrella term that encompasses several different types of AI models in use today, and new types of models being refined or developed all the time. The big, recent breakthroughs have been in Large Language Models (LLM), derived from deep neural network advancements in 2012. This initial breakthrough was related to image classification, not text, but applying the same techniques to language learning created much of what has driven the explosion in Generative AI.
Basically, this is a technology that allows chatbots to be much smarter and more capable. Many people already use them like they would a Google search, or as a sounding board for ideas, or a personal assistant to draft or summarize emails, memos, and other documentation, or organize other facets of personal or professional life. It’s capable of a lot more already, though. And capabilities are growing, seemingly daily.
More Around the Corner
AI technology is not plateauing. It’s advancing and evolving at a rapid pace. Ironically, computer vision is a quietly advancing branch of AI that goes back to the initial goals of the neural networks, allowing computers to “see” and process visual information in ways that match or exceed humans.
Agentic AI (often just referred to as an AI Agent) is also an emerging field in AI tools, where the AI can execute defined tasks independently. If you ask it to book a flight, it can find you the best price and book it with little more than a short prompt.
There are also AI tools trying to package several or all types of AI tools into one interface, often referred to as Multimodal AI. This would fully realize the dream of AI personal assistants for everyone, able to complete a broad variety of complex tasks with simple, intuitive commands.
How Smart is it?
As impressive as AI tools can be, they are not alive or sentient. Fundamentally, the Large Language Models (LLM) work a lot like the auto-text predictor on your phone has for the past several years. LLMs just predict a lot more words in a row and have a much larger dataset with a lot more sophistication behind them. But fundamentally, they each work by predicting the next best word to use. Even if it sometimes seems smarter than your employees, or seems like a better conversationalist, it’s not alive. AI doesn’t have ideas of its own.
But, it can synthesize and summarize all the ideas people in similar positions have shared, and it can match which of those seems most like whatever situation you describe. There are AI tools that can analyze big datasets and draw insightful conclusions or present options most humans might otherwise miss.
Should I Implement AI at my Business?
Most CEOs feel one of two ways about AI—anxious and excited, or annoyed and resistant.
Some CEOs can’t wait to adopt this new technology. They might see it as the cure to all their problems, the key to explosive overnight revenue growth, a way to reduce labor costs, or the foundation of a new billion-dollar business. If you feel this way, it’s probably time to take a breath and slow down. There’s a lot of AI snake oil out there, and your eagerness can make you susceptible to shelling out a lot of money for a solution that’s obsolete before it’s implemented, if it works at all.
Other CEOs might feel great about their current state processes and fundamentals. You probably don’t see a reason to invest time, effort, and money in a technology that seems like a fad, or that doesn’t necessarily apply to your business. If you feel this way, you’re in danger of being left behind. AI has the potential to scale operations and reduce cost in a way you just can’t compete with if all your competitors implement successfully ahead of you.
On AI Integration
“The most dangerous thing I hear from executives is, ‘We need AI to stay competitive.’ That’s not a strategy; it’s just calling for artillery without a target. The winners in business build a mission-oriented strategy, not one based in a fear of missing out. They ask, ‘What is our combined capability, and which specific problem will we task AI to solve?’ That clarity is the difference between a successful operation and an expensive science experiment.”
–James Suh, Marine Corps & US Navy Veteran
Founder & CEO of Nashville Analytics
So, What Should I Do?
Find some ground in the middle. While not true for all industries or companies, for most small businesses, this is a time to learn more about what AI technology does, how it works, and which of your problems it could solve. This isn’t something to just leap into or randomly experiment with. This is something to consider carefully, and evaluate with a structured approach to arrive at the best choice.
How to Audit Your Company for AI Use-Cases
To evaluate the potential for AI tools and services at your company, you really need to do the work. This isn’t the time to go with your gut or engage in expensive experimentation. It deserves a systematic approach to making a good decision. The eight steps that follow are going to give you the best chance of adopting AI in a way that’s going to lead to a positive experience for you, your company, your employees, and your customers.
Step 1: Talk To Your Employees, Get Their Buy-In
A majority of workers are worried about AI taking their jobs. Don’t let rumors bring the news ahead of you. Inform everyone you’re performing an AI use-case audit of the business. If you want this process to be successful, you need your employees to believe that AI can make the business better, and your vision for AI adoption includes them, and is aimed to improve their day-to-day experience at work. This means everyone needs to be involved. All your employees, from the most junior people on the organization chart all the way up to the leadership team.
If you don’t get that buy-in, employees won’t be open and honest about potential use cases that overlap with their job responsibilities. They’re going to be afraid and go into those conversations trying to protect their job. That sabotages the potential of this entire process before it even begins.
While talking to your employees is listed here first, they really should be involved at every single step along the way. AI adoption for small business generally isn’t a top-down process, it’s more bottom-up, integrating at the front lines of your business.
Surveying your employees after your initial announcement can help give you a sense of where they’re at. We like the sample survey Lattice has built as a starting point.
Step 2: List Business Pain and Friction Points
There are parts of your business that are tedious and frustrating. There are processes that you’ve banged your head against since launching that are still not working. Some processes might be slow in a way that dampens customer enthusiasm, or worse, loses them altogether. There are parts of business that lean on making the best choice with the information available that are imperfect. Others might be labor intensive or tedious, keeping your talent away from revenue-generating activity that grows business.
Talk to your frontline employees, figure out what they spend the bulk of their time doing, or waiting to do. Ask employees what task they’d love to get off their desk so they can focus on what drives revenue, services customers, or moves business forward.
Don’t forget about your customers, either! Their pain and friction points can be just as useful to reveal areas where AI technology can help elevate your customer experience. Offer a survey, perhaps for store credit or a small discount to encourage participation for those who had a typical or good experience. You shouldn’t even mention AI here, unless its to gauge customer feelings about AI usage in general.
Once you’ve done what feels like a complete review of business, you should have a giant list of pain points and friction points that are slowing down business and business growth. Take some time to identify some of the biggest ones holding you back right now. It might be a long list.
When you’re going through the list and trying to decide which problems need to get solved the most urgently, keep in mind AI isn’t going to be the answer for every single one. You might be able to make some very simple changes to fix problems that same afternoon. Don’t dismiss obvious opportunities to improve business the old-fashioned way just because the primary purpose of the review was to look for AI integration opportunities. Just because AI technology exists doesn’t mean you abandon common sense about needing a new hire, needing to fix a broken process, refreshing an employee’s training, or pivoting toward what customers want.
Fix the easy stuff, prioritize what’s left.
Step 3: Check Digital Readiness
AI runs on data. Does your entire business still run on sticky notes, legal pads, and giant filing cabinets (or worse, out of your head)? AI might be limited in how it can help you. AI needs digital data input before it can start to give you the best, actionable output.
If bad or incomplete data goes into the AI tool, it’s going to degrade the quality of results it puts out.
Even if all your data is in some sort of digital format, there are still some red flags that are going to make integration much more difficult. Is your digital data:
- Scanned PDFs or photos of text, or handwritten text?
- Random word docs of data that should be in a formal database?
- In databases, but column and row naming conventions and order is inconsistent?
- In a proprietary file format that requires a special or bespoke tool for AI to parse?
- Untagged data, no meta-data?
Each of these red flags can make the data less accessible for AI tools to ingest and utilize. If your data doesn’t have any red flags, and uses industry-standard formats that already integrate with most cloud-based tools and APIs, you should be set up well for AI tool adoption in the current climate.
If not? When you get to the end of this step, you might look at making some changes to your data formatting and storage, or even hire someone to update archive data to the current standard. It’s best to consult with an expert ahead of making big changes to ensure you’re being smart about what you change and how.
You’ll also want to think about data privacy and compliance. A misstep on data privacy can be a disaster for your business. Make sure you’re thinking about this now, so you can ask questions about it in future steps.
Step 4: Build AI Literacy Org-Wide
Now that you’ve made some evaluations, you’re ready to learn what AI tools can do. There are several approaches to this. You might bring in consultants, integration strategists, speakers, and workshops built around teaching AI use-cases to educate your entire team in a series of events.
There are non-profit and industry groups offering education. There are even executive AI programs at MIT, Kellogg, Oxford, and other major institutions where you might gain some formal education about the current AI integration landscape, though these can be expensive and are probably overkill. Many AI tool vendors are also often willing to do some of this education, but of course there’s a natural bias toward selling their tools to solve your problems.
The important thing is to educate everyone in the organization to some degree about the power and pitfalls of AI tools, and give them a high-level education about the types of tools that are out there and what they can do. During these educational opportunities, employees should be learning with an eye toward possible use-cases that might increase their speed, precision, or add entirely new capabilities in their role. Collect these ideas from employees after educational sessions to use in the next step.
Step 5: Match AI Tools to Pain/Friction Points
So, now you have a list of problems you want to solve, and you’ve learned enough about AI tools to understand what the technology can and can’t do. Look at your biggest problems, and match them to AI tools. Are your analytics and projections not accurate enough? Predictive modeling AI can address that problem so you can better budget around realistic sales projections and make more accurate inventory orders. Is your product quality control poor? Computer vision AI can improve anomaly detection.
If you haven’t already, make some basic matches between problems and use-cases. This is going to help you research basic cost estimates for implementation, so you can meet with vendors coming from a place of preparedness and knowledge.
Step 6: Check In and Test
You probably have a good idea at this point what you want to do. Check in with your employees, see what, if any, reservations they might have about some of your top solutions. They might be unfounded fears, or they might be something you really need to flag when you work with a vendor to prevent disaster.
Check in with your customers. Float these ideas as hypotheticals with a few trusted clients and gauge their reaction. It’s possible your customers are going to overwhelmingly reject your AI use-case on some moral or ethical ground, or just because of the way the perception of that use-case is trending. It’s best for you to know about any such headwinds before you cut a check to a vendor.
Another area to reconsider is data privacy. If you adopt an AI tool, will your data be safe, secure, and in compliance? Do you have customer consent to feed this data into a tool? Is it necessary to get consent before moving forward?
This is also a time where you might test pilot a use-case if there are cheap or free commercial options out there to play with, even if they’re not going to be quite as integrated or effective as a customized vendor tool might be. A pilot can give you an opportunity to trial a new process, and put the AI-integrated process up against your existing process head-to-head. Just keep it very low cost, and small scale.
However possible, you want to compare apples to apples in any test. How did the test go compared to the same time frame last year or last month? Have two employees go head-to-head with one using the new tool and one using the traditional way for a set time frame. Compare the results to each other and historical performance. Does it feel like there’s something promising in the tool?
Step 7: Consider Costs and Risks
You should feel very prepared at this point. Research the cost of products and implementation for similar use-cases at similar scales. Weigh the risks based on what you’ve learned talking to your frontline employees and your customers. Ask yourself these questions:
- Is this AI tool integration likely to address the pain point in a way that is worth the cost? Especially considering that this is still emerging technology likely to improve and change drastically in the short term. Will this solution be obsolete in a year?
- How are my current customers and clients going to feel about this? Will they cheer it, leave for someone not using AI, or unlikely to notice any improvements on their end?
- Is an AI tool the only or best way to solve this pain or friction point? Revisiting a process or making a key hire can still be a lot cheaper and more effective than an entire AI system.
This technology is very exciting and new right now. However, if it’s not the ideal solution to your problem and it’s a dollar amount that makes you flinch, it’s probably best to hold off. This technology is advancing very quickly. Unless this is going to significantly improve your business today, or give you a huge market advantage, it might not be worth implementing yet. In just a few short years, all-in-one AI solutions are going to be smarter and more widespread, and the cost is likely going to continue dropping.
Step 8: Choose a Vendor or Make a Hire
You’re still all in? Then it’s time. Research vendors or have HR start looking for someone capable of working in-house to manage AI tool integration long term. Use the same rigor you would to hire any other vendor or add a new tech position. Hear pitches, interview multiple candidates, the works. You’ve put this much time into the process, don’t waste it by rushing this step.
As you go through this process, you’ll want someone in-house that is responsible for the AI tool. AI tools can drift off-course and models can break if no one is monitoring them. This employee can be responsible for ensuring the tool remains updated as the business grows or processes change. They’ll also help ensure new hires are trained appropriately to make best use of the new tools.
Finally, before you sign a contract, make sure you understand very clearly if the vendor owns or is allowed to train their own models on your data. Ensure your IP is protected, who can access the model, and who can export your data, even if you change tools or vendors down the road.
Don’t Go Through AI Integration Alone
Join a nationwide community of veteran and military spouse entrepreneurs facing the same challenges. IVMF has a huge portfolio of entrepreneurship programming designed to meet veterans and military spouses like you precisely where you are on your journey, no matter if you already have a small business empire, or just an idea on a napkin. Learn more about our diverse portfolio of entrepreneurship program offerings today!