THE SHIFT: Business intel for those who don’t sit behind a desk.

How Small Businesses Can Use AI to Compete With Bigger Companies

Big companies have departments.

They have marketing teams, analysts, developers, designers, lawyers, accountants, customer service departments, and enough people to hold a meeting about the next meeting.

Small-business owners often have themselves, a few trusted employees, and a phone that never stops vibrating.

That is why the conversation about artificial intelligence feels strangely backwards. AI is often discussed as a threat to small businesses and their employees. But large companies are already investing in customized AI systems, training teams to use them, and incorporating the technology into everything from advertising to logistics.

Small businesses are the ones being told to wait, worry, and hope the whole thing goes away.

We do not think it will.

For an independent business competing against companies with more money, more people, and better technology, refusing to use AI may be a luxury it cannot afford.

That does not mean trusting a chatbot to run the company. It does not mean firing the photographer, ignoring the accountant, or publishing the first thing a machine produces.

It means using a new tool to gain capabilities your business could not previously afford.

AI is not only for companies with innovation departments

The usual stories about AI in business involve enormous restaurant chains, hotel groups, banks, and technology companies. Those examples demonstrate what the technology can do, but they can feel irrelevant to an owner trying to answer customer emails while checking inventory and covering an employee’s shift.

Small businesses usually have different reasons for using AI.

They are not trying to automate a global operation. They are trying to recover an hour, finish a neglected project, understand a report, create a piece of content, or solve a problem without adding another monthly subscription.

Research suggests that this is already happening. An OECD survey found generative AI in use at approximately 31% of the small and medium-sized businesses it studied. Among the businesses using it, 65% said it had improved employee performance. Only 29% said they were using it in the company's core activities.

In other words, most are not handing the keys to a machine. They are using it around the edges, where administrative work, research, communication, and unfinished tasks quietly consume the day. By OECD

That is close to what we have experienced at Under NY Sky.

We are a workwear and apron company, not an AI startup. We did not begin using AI because we wanted to become a technology company. We began using it because the list of things a growing business should do was much larger than the team and budget available to do them.

What using AI actually looks like at a small business

The phrase “using AI” can make it sound as though a company installed one magical system and automated the building.

The reality is far less cinematic and far more useful.

At Under NY Sky, we use different AI tools to assist with advertising control, website development, customer service, research, images, videos, translations, content, employee materials, international logistics, and preliminary administrative questions.

Everything begins with us.

We identify the problem. We provide the information. We evaluate the output. We correct what is wrong. A person decides whether the final result is good enough to use.

AI has not taken control of the business. It has helped us regain control over work that previously depended on limited time, outside schedules, or budgets we did not have.

A website change should not always require weeks in line

Small companies depend heavily on outside specialists. Those relationships are valuable, but they can also make minor changes surprisingly slow.

A developer may be serving several clients. A small website update can be queued behind larger, more profitable projects. The change itself might require only a short amount of work, but the business may wait days or weeks for someone to reach it.

AI-assisted development allows us to handle more of those updates quickly. It has also helped us create a tool to control advertising and improve other parts of our platform.

This does not mean every experiment works. We have started building internal applications only to discover that the idea was unnecessary, impractical, or not good enough to deploy.

Failure is part of the process. The difference is that testing an idea has become faster and more accessible.

Other small-business owners are seeing something similar. Tony Mistry of Houston’s BnX Burgers once built employee schedules manually before moving to scheduling software that cost $68 per month. He later began using ChatGPT to create updated schedules from previous schedules and changes in employee availability. According to theHouston Chronicle, the process now takes minutes.

Mistry has also experimented with using AI for bookkeeping reports. He still has to correct errors, which is not a minor detail. But reports that once took at least 15 days can now be prepared much faster and at a lower cost. Read on Houston Chronicle

The lesson is not that every owner should cancel scheduling or bookkeeping software tomorrow.

The lesson is that small businesses can now question whether every operational problem requires another subscription, another platform, or another long wait.

Sometimes the right answer will still be yes. Sometimes a specialized product is safer, better, and cheaper than maintaining an internal tool. But now owners have another option.

AI can add a capability without replacing a person

One of the greatest fears around AI is that every use of it represents a job eliminated.

That has not been our experience.

Most of the lifestyle images on our website are now produced with the help of AI. Before, we could not afford to update that type of content as frequently as an e-commerce business needs.

Creating one traditional campaign might involve a studio, models, a photographer, a stylist, samples, schedules, and weeks of planning. Getting a new product onto a model could take one or two months.

Today, we can sometimes complete that process in approximately one week and for a fraction of the cost.

But there is an important part of the story that is easy to miss: our products are still photographed by a professional photographer. Those real photographs provide an accurate foundation. Our designer then uses AI as part of a larger creative process.

We still work with the same photographer and the same designer. They can simply deliver more than they could before.

The same applies to promotional videos. A video that previously took months to complete can now be ready in approximately 15 days. AI did not replace a video department. We never had a large video department to replace. It gave our existing team a capability the business largely lacked.

A family-owned restaurant in Los Angeles offers an even more dramatic example. Christian Ortega, who handles marketing for The Original Tamale Company, used ChatGPT to assist with the script for a humorous social media advertisement. AI also contributed to its narration.

He reportedly made the video in ten minutes.

Within about three weeks, it had received more than 22 million views and 1.2 million likes. Customers told the restaurant that the video was the reason they visited. Read on Business Insider

AI did not build the family’s reputation, make the tamales, recognize the company’s personality, or guarantee that millions of people would care. Ortega noticed a trend, understood his audience, and adapted the idea to the business.

The machine accelerated execution. The human supplied the taste.

Small businesses do not always need lower costs. They need greater output.

We currently spend more on technology than we did before adopting AI.

That may sound like a terrible endorsement, but it is one of the most important points in this article.

AI has not eliminated our subscriptions. It has added new ones. The difference is that the business receives far more in return.

The real calculation is not:

Did our software bill go down?

It is:

How much more can the same team accomplish with the money and time available?

For a small ecommerce company, saving thousands of dollars on image production matters. Reducing a project from two months to one week may matter even more. Products, seasons, advertising opportunities, and customer expectations do not politely wait for a small business to catch up.

Risa and David Baron, owners of Jackie’s Jams in California, offer a smaller but equally revealing example. Risa uses generative AI for advertising messages and back-office work, and estimates that it saves her at least one hour a day. Read onU.S. Chamber of Commerce

One hour does not sound revolutionary.

Over 250 working days, it becomes 250 hours. That is more than six 40-hour workweeks returned to the owner.

Small businesses are not always looking for a revolution. Sometimes they need Tuesday afternoon back.

Let AI prepare the work, not make the final decision

AI can be extremely helpful with subjects that small-business owners cannot navigate on a day-to-day basis.

We use it as a preliminary assistant for straightforward legal and administrative questions. It helps us understand unfamiliar terminology, interpret international logistics documents, and determine what information a form is requesting. It can help us organize a question before contacting an accountant or lawyer.

That last part matters.

AI is not our lawyer or accountant. It does not assume professional responsibility when an answer is wrong. It may overlook a recent regulatory change, misunderstand the circumstances, or confidently invent information.

For consequential legal, tax, employment, safety, or financial decisions, preparation and explanation are not substitutes for qualified professional advice.

Used properly, however, AI can make those professional relationships more efficient. An owner can arrive with organized documents, a clearer understanding of the issue, and better questions.

Mike Salvatore, who owns two cafés, two bars, and a bicycle shop in Chicago, uses ChatGPT to analyze food costs. Work he once completed once or twice a year can now be revisited every three weeks. He described AI as essentially serving as his CFO onThe Wall Street Journal

We would add one word to that description: preliminary.

AI can help an owner see patterns and make decisions. A responsible human must still understand the data, question the conclusion, and own the result.

Customer service is a good place for boundaries

Our website now has an AI-assisted bot that answers common questions. It can help customers find basic information without waiting for someone to become available.

It also sends important questions to a person.

That handoff is not evidence that the technology failed. It is evidence that the system was designed around the customer rather than around an automation target.

A bot may be excellent at answering a frequently asked question about shipping or at directing someone to the correct page. It should not improvise when a customer has an unusual product issue, a sensitive complaint, or a question with significant consequences.

Hospitality businesses are applying the same principle. Some restaurants use AI to answer phones or manage routine reservations so employees can focus on the guest already standing in front of them. The objective should not be to remove hospitality from the experience. It should be to stop repetitive tasks from pulling people away from it.

The best test is not “Can AI do this?”

It is “Will using AI here improve the customer’s experience, or merely make our operation cheaper?”

If the customer receives worse information, fewer choices, or no path to a real person, the business has automated the wrong thing.

AI can help create content, but it cannot care about your reputation

At Under NY Sky, we still write the substance of our product descriptions. We know the products, materials, uses, strengths, and limitations.

AI helps us refine that information for traditional search engines and the growing number of answer engines that interpret product content. It can suggest clearer structures, identify unanswered questions, assist with translations, and adapt material into captions or other formats.

But publishing unreviewed AI content is one of the fastest ways to make a real business sound like every imaginary business on the internet.

Reddit discussions among small-business owners reflect this conflict. One commenter said that most AI tools created more trouble than they were worth, but that ChatGPT had been useful for product descriptions and Canva’s AI features saved time on social graphics.Reddit Community

Another small-business discussion included a much harsher warning: AI can be confidently and wildly incorrect, so its work must be verified.Reddit Community

Both perspectives can be true.

AI can save hours and still produce bad work. Speed and reliability are not the same thing.

In our own experience, more than 90% of raw AI output contains something we would change, and much of it is rejected completely. Images arrive with incorrect product details. The text sounds generic. Research misses context. Code may solve one problem while creating another.

The first answer is rarely the finished work.

That does not make the tool useless. It makes supervision part of the job.

Treat AI like an intern, not an oracle

AI is frequently marketed as a senior professional who knows everything, works instantly, and never takes a day off.

That is a dangerous way to use it.

A better comparison is an extremely fast intern who has read an astonishing amount, can attempt almost anything, and occasionally delivers an answer with total confidence despite having misunderstood the assignment.

You would not give a new intern an unexplained task, publish the result without reading it, and then blame the intern when the business suffered.

You would provide context. You would explain the goal. You would show examples. You would review the work. You would correct mistakes. You would keep final responsibility with someone who understands the business.

AI requires the same discipline.

The OECD found that, in one survey, half of small and medium-sized businesses said their employees lacked the skills needed to use generative AI. That points to an uncomfortable truth: paying for the tool is easy. Learning how to direct it, evaluate it, and fit it into a real workflow is the harder investment.OECD paper

The companies that benefit most will not necessarily be the ones with the longest list of AI subscriptions. They will be the ones who understand their own work well enough to know where assistance is valuable.

Do not feed the machine everything

Useful AI often needs context, but context can include information that should never be casually uploaded to an outside platform.

Customer records, employee information, unreleased financial results, contracts, passwords, private communications, bank statements, and proprietary business information deserve special care.

Before using any AI service, owners should understand what data the tool collects, how that data may be retained, who can access it, and whether it may be used to improve future systems. Sensitive information should be removed or anonymized whenever possible.

Official cybersecurity guidance for small businesses specifically identifies data leaks, privacy breaches, unreliable output, manipulation, and supply-chain vulnerabilities as AI-related risks. It warns that uploading customer or staff information without proper anonymization can expose private data. Read on Australian Cyber Security Centre

Privacy regulators have also advised organizations to clearly identify public-facing AI systems, such as chatbots, and to consider how human oversight will be incorporated. read onOffice of the Australian Information Commissioner

You do not need a 70-page AI policy before asking for help with an Instagram caption. You do need rules before employees begin uploading private business information wherever they please.

Begin with the work that is not getting done

A small business does not need an AI strategy retreat.

It needs an honest list.

What work repeatedly causes delays? What information is difficult to understand? What task is consuming hours without creating much value? What project has remained on the wish list because the company cannot afford it? What customer question is answered 20 times every week? What paid application solves a problem simple enough that an internal tool might work?

Choose one.

Start with a low-risk task where a mistake can be caught before it reaches a customer or affects an important decision.

Good starting points may include:

• Turning meeting notes into an organized first draft
• Rewriting an original product description for clarity
• Preparing interview questions or a job-posting draft
• Converting existing company information into an employee guide
• Categorizing customer comments into recurring themes
• Translating internal material for review by a fluent speaker
• Creating a first version of a spreadsheet formula
• Comparing the features and costs of software subscriptions
• Drafting captions from content the business has already created
• Identifying repeated customer questions for an FAQ
• Building a prototype of a simple internal calculator or tracker

Define what success means before beginning. It might be hours saved, faster response times, fewer missed questions, lower production costs, or a project finally completed.

Then check the result.

If the AI creates more work than it removes, stop using it for that task. We have done that too.

AI does not deserve permanent employment merely because it is fashionable.

The bigger companies are not waiting

Independent businesses have always competed with companies that possess advantages they cannot duplicate.

AI will not erase that inequality. Large companies can build private systems, hire specialized teams, buy enormous datasets, and integrate tools across entire operations.

But for the first time, a small-business owner can access a modest version of capabilities that once required several departments.

An independent restaurant can analyze customer feedback against its brand message. A shop can create a simple internal tool. A service business can turn years of knowledge into training material. An e-commerce company can produce campaigns it previously could not afford. An owner can understand a report before entering a meeting with an accountant.

Restaurant Dive recently asked whether AI could help independent restaurants close the data gap with large chains. That is the right question. Independents cannot copy the technology infrastructure of a national chain, but accessible AI tools can help them examine their own information and make better use of what they already know. More on Restaurant Dive

Small businesses should not use AI because technology companies say they should.

They should use it because their competitors are not standing still.

At Under NY Sky, AI has not replaced the people who understand our products, customers, and identity. It has not removed the need for professional photography, design, development, legal advice, accounting, or human judgment.

It has allowed the same people to move faster, test more ideas, and complete work the business could not previously afford.

AI is imperfect. It can waste time, invent facts, produce embarrassing images, expose sensitive information, and create a remarkably polished version of the wrong answer.

Use it anyway.

Use it carefully. Use it skeptically. Give it boundaries. Check its work. Keep people responsible for the decisions that matter.

While a small business is deciding whether AI is good or bad, the larger company down the road is deciding where to use it next.

And choosing not to participate may be the most expensive choice of all.