Putting AI to Work Without Losing the Human Side of Business

AI is not coming someday.

It is already sitting inside our organizations, helping people write faster, analyze information, build models, organize processes, and complete work that used to take hours.

The real question is no longer whether businesses will use AI.

The question is whether we will use it thoughtfully.

That was the focus of our recent Ask the CFO conversation with Dolores Hirschmann, founder of Masters in Clarity. Dolores has spent her career at the intersection of technology, marketing, strategy, and human behavior. She has also been helping me understand what AI can actually do beyond writing an email or cleaning up a paragraph.

What stood out to me during this conversation was that AI is not simply another piece of software. Used correctly, it can change how work moves through an organization, how teams spend their time, and how leaders think about productivity, staffing, and growth.

But getting there requires more than opening ChatGPT and typing a question.

Start With the Work You Do Not Want to Do

Dolores offered a very practical place to begin.

Look at your week and identify the work that consumes too much time, gets repeatedly postponed, or feels unnecessarily tedious.

What is the task you keep avoiding?

What takes three hours but should take thirty minutes?

What work is repetitive, transactional, or administrative?

Then ask whether AI could handle 80% of it while you remain responsible for the final 20%.

That is a much more useful starting point than trying to learn every new AI platform at once.

For me, AI has already made a meaningful difference in how I prepare for meetings. I can take meeting transcripts, turn them into summaries, develop follow-up points, and create agendas for the next conversation. Work that once required additional administrative time can now happen much more quickly.

The same principle can be applied to proposals, job descriptions, policies, financial models, operating procedures, research, and recurring reports.

You do not have to automate the entire job.

Start by removing the part of the job that should not require so much of your time.

Stop Looking for One Tool That Does Everything

One of the traps people fall into is asking which AI platform is the best.

ChatGPT, Claude, Perplexity, Copilot, and other tools all have different strengths. There is no single answer that applies to every company or every task.

Dolores described ChatGPT as a strong thinking partner. It can help organize ideas, develop content, structure processes, test your thinking, and provide another perspective.

She described Claude as more of an action-taking team member—something that can work with larger amounts of information, organize files, analyze data, create outputs, and help move a project forward.

Other tools may be better suited for research, meeting notes, team communication, or a very specific operational need.

The important point is not to spend six months evaluating tools while accomplishing nothing.

Pick one.

Learn how it works.

Use it until it becomes part of your normal process.

Then expand when you have a genuine business reason to do so.

We are moving from what Dolores described as the horse-drawn carriage to the first automobile. At this stage, arguing over which model is the Ferrari can become a distraction.

Almost any thoughtful use of today’s tools represents a major step forward from what was available only a few years ago.

AI Becomes More Valuable When It Understands Your Business

The next level of AI adoption is not simply asking isolated questions in a chat window.

It is giving the system useful context.

That might include meeting notes, project documents, financial information, approved messaging, operating procedures, or information stored within tools such as Gmail, Google Drive, Teams, or Slack.

With the right permissions and safeguards, AI can begin to function less like a search box and more like a team member that understands the project.

Dolores shared examples of using AI to help evaluate a potential acquisition, organize financial and operational data, project future performance, and compare written information with notes from previous conversations.

She also described using AI alongside a difficult CRM system. Rather than handing over complete control, she took screenshots when she became stuck and asked the system to guide her step by step.

That distinction matters.

AI adoption does not have to begin with full automation.

It can begin as coaching.

It can tell an employee where to click, help interpret a report, explain an error, or guide someone through a process they only perform occasionally.

There is a wide range between doing everything manually and turning over complete control. Every business has to find the level of access and autonomy that fits its risk tolerance.

Security Cannot Be an Afterthought

One of the most important questions in the session came from a finance leader who raised concerns about privacy, data access, and confidential information.

Those concerns are legitimate.

Businesses should not casually load sensitive financial, employee, customer, health, or strategic information into free public tools without understanding the platform’s data policies.

Paid business, team, and enterprise accounts generally provide stronger administrative controls and privacy protections, but leaders still need to understand exactly what they are purchasing and what information employees are allowed to share.

Permissions also matter.

When AI is connected to folders, email, communication platforms, or business software, access should be deliberate and limited. Employees should know what the tool can see, what actions it is allowed to take, and when human approval is required.

AI can move quickly.

That is useful when it is working within the proper guardrails and dangerous when it is not.

The cost of a business account is small compared with the possible cost of mishandling sensitive information.

The Objective Is Not Simply to Eliminate People

The conversation eventually came back to something I have believed throughout my career.

Whenever I entered a new organization, I tried to understand how much of the finance team’s time was being spent on transactional work and how much was being spent on analysis, strategy, and value creation.

Sometimes the balance was 99% transactional and 1% analytical.

Sometimes it was closer to 80/20.

My goal was never simply to eliminate people. It was to improve that ratio.

AI gives organizations a new opportunity to do that.

If a financial professional is spending hours assembling a recurring report, AI may help create it faster. That professional can then use more of the day to understand what the numbers mean, identify risks, ask better questions, and advise leadership.

If an HR executive is spending weeks assembling policies or rewriting job descriptions, AI can help create a strong first draft. The executive can spend more time evaluating workforce needs, supporting managers, strengthening succession plans, and addressing employee concerns.

If a manager is repeatedly gathering updates from multiple employees, an AI tool may help organize those updates into a useful summary. The manager can spend more time removing obstacles and developing the team.

That is the opportunity.

Not humans or AI.

Humans working differently because AI can carry more of the repetitive load.

Experienced Professionals May Be More Important Than Ever

There is a common fear that AI makes experience less valuable.

I believe the opposite may be true.

AI can produce an answer, but someone still has to determine whether the answer is good.

Experienced professionals understand context. They recognize when assumptions are unrealistic, when a number does not make sense, when a policy creates unintended consequences, or when a recommendation ignores how the business actually works.

The people who learned accounting before sophisticated systems existed could look at a spreadsheet and recognize a bad output.

The people who have built companies, managed teams, navigated downturns, and solved difficult problems are the people best equipped to judge AI-generated work.

The opportunity for experienced leaders is to position themselves as experts who know how to work alongside AI.

They do not need to become software engineers.

They do need to remain curious enough to understand what the tools can do and confident enough to challenge the output when it is wrong.

Adoption Is a Leadership and Culture Question

Technology may be the easiest part of AI adoption.

The harder part is people.

Some employees are afraid AI will take their jobs.

Some managers do not want to reveal that a task can be completed faster.

Some established employees are comfortable with the way work has always been done.

Some organizations are so concerned about risk that they restrict the technology until it is almost impossible to demonstrate value.

Leaders have to address those concerns directly.

Telling employees to “use AI” is not a strategy.

Employees need to understand why the organization is adopting it, which tools are approved, what information is protected, how success will be evaluated, and what the technology means for their role.

Dolores shared an example of a large company that gave groups of interns controlled AI projects and then asked them to present what they learned to company managers.

I thought that was a smart approach.

The organization created a sandbox. It encouraged experimentation without immediately exposing the company to unnecessary risk. It also allowed newer employees to show established leaders what might be possible.

Companies do not need to have every answer before they begin.

They do need to create a responsible environment where people can learn.

Start Small, but Start

AI can feel exciting, frustrating, overwhelming, and distracting—sometimes all in the same day.

You can spend an enormous amount of time trying new platforms, comparing outputs, and following every new feature announcement.

Curiosity matters, but it should be connected to a real outcome.

Choose one recurring frustration.

Choose one administrative bottleneck.

Choose one task your team completes every week.

Then test whether AI can make it faster, clearer, or more accurate.

Measure the result.

Keep a person involved.

Protect the information.

Document what works.

Then move to the next process.

That is how AI becomes useful—not as a collection of impressive demonstrations, but as part of the way your organization operates.

The companies that succeed will not necessarily be the companies that purchase the most tools.

They will be the companies that know which work should be automated, which decisions must remain human, and how to help their people become more valuable as the technology improves.

AI may not give every business owner Friday off immediately.

But used thoughtfully, it can give people back the time they need to think, lead, solve problems, and do the work that only humans can do.