PUT AI TO WORK

Start With One AI Workflow

A practical way to test AI in your business without taking on a giant project.

Issue #12 · Wednesday, September 9, 2026

Small business AI use has nearly doubled since 2023 — from roughly 30% to 60% today, according to the U.S. Chamber of Commerce's latest small business technology survey. And among small businesses already using it, 82% grew their workforce over the past year, not shrank it. AI is not replacing the team. It is changing what one person can get through in a day.

You do not need an AI strategy for your entire business. You need one task that happens often, follows a recognizable pattern, and takes more time than it should.

Start there. A small, measurable test will tell you more than a long list of tools or a broad plan ever will.

In This Issue

• Choose the task before the tool

• Run a two-week AI test

• Decide whether to expand

This issue contains affiliate links. If you purchase through these links we may earn a commission at no additional cost to you.

Choose the task before the tool

The easiest mistake is starting with a product. You see a new AI agent, imagine everything it might do, and then spend your afternoon searching for a place to use it.

Reverse that order. Start with a task you already understand. Good candidates include coordinating and rescheduling appointments, sorting and tagging inbound orders, drafting social media captions, processing expense reports, or organizing new-hire paperwork.

Look for something that happens at least a few times each week. It should have a clear starting point and a clear definition of "done." You should also be able to review the result before it reaches a customer, colleague, or financial system.

That last point matters. AI is useful when it removes repetitive effort while leaving you in control of the final decision. It is less useful when a mistake creates a customer problem that takes longer to fix than the original task.

Try this selection test: Can you describe the task in one sentence? Does it repeat often enough to matter? Can you show the tool a few examples of a good result? And can you check the output quickly?

If the answer is yes to all four, you have a reasonable first test. You do not need to automate the whole process. Begin with the part that consumes the most attention.

Run a two-week AI test

A short test keeps the decision grounded. Instead of asking whether AI could transform your business, ask a smaller question: does this tool make this one task faster or easier without lowering the quality of the result?

First, write down your current baseline. How long does the task usually take? How many cases do you handle in a normal week? How much editing does the work require? You do not need a perfect measurement. A rough starting point is enough.

Next, choose one tool and keep the test narrow. Claude can help draft and organize work from written instructions. Taskade can help coordinate projects and AI-supported work in one place. Make can connect steps across business apps when the task needs information to move from one system to another.

The right choice depends on the task. If you need a first draft, start with a tool that can work from examples and clear instructions. If you need information to move between apps, look at a connector such as Make. If you need a shared place to organize the work, Taskade may be a better fit.

For the first two weeks, limit the test to a small, known set of cases. Keep the original work available so you can compare the result. Review every output where accuracy, money, privacy, or a customer relationship is involved.

Measure one outcome. You might track minutes saved, response speed, completed tasks, or the amount of editing required. One useful measure is better than a dashboard full of numbers you never check.

Decide whether to expand

At the end of the test, do not judge the tool by how impressive the first demonstration looked. Judge it by the work it helped you complete.

Continue when the result is repeatable, useful, and easy enough to review. For example, an AI draft may not be ready to send, but it could still be valuable if it gives you a strong starting point and cuts the writing time in half.

Change the task or stop when correction time erases the benefit. If every output needs a complete rewrite, the task may be too ambiguous. If the tool works only with extensive preparation, choose a simpler use case. A failed test is useful when it tells you what not to repeat.

If the test works, expand one step at a time. Add more examples, try a slightly larger batch, or connect one additional business app. Keep the original measure so you can tell whether the improvement continues as the workload grows.

The goal is not to add AI to every corner of your business. The goal is to find one place where it earns its keep.

Keep testing practical use cases, one at a time, and let the results guide what comes next.

That's Issue #12.

More practical ways to put AI to work are ahead. Until then, choose one task, run a small test, and pay attention to the result.

Know someone who is curious about AI but unsure where to begin? Forward this issue to them.

— Put AI to Work