AI Can Help You Build a Budget. It Cannot Replace the Process

Blog

AI Can Help You Build a Budget. It Cannot Replace the Process

Artificial intelligence has changed the way we work.

Today, a business owner can ask ChatGPT to develop a marketing plan, draft a proposal, create a project budget, suggest a ticket price, prepare a communication strategy or even estimate what it might cost to put an event together.

And, to be fair, AI can do a remarkably good job.

But there is a problem I am increasingly noticing: we are beginning to confuse an AI-generated answer with an informed decision.

They are not the same thing.

AI can help you think. It can help you structure information. It can help you identify things you may have forgotten.

But it cannot magically know the price your supplier quoted you yesterday, how much transport will actually cost, what your venue is charging, how many people you need to employ, what your permits will cost, how much equipment you need, or what unexpected expenses your particular project will encounter.

That information still has to come from somewhere.

And that is where the process matters.

A budget is not a guess

Consider an event.

You ask AI:

“What would it cost to organise an event for 300 people in Mombasa?”

The system might give you a beautifully structured budget.

Venue: X.

Catering: X.

Entertainment: X.

Marketing: X.

Transport: X.

Production: X.

Contingency: X.

Total: KSh 841,000.

It looks professional.

It has headings.

It has numbers.

It may even have a nice table.

But there is a fundamental question we need to ask:

Where did those numbers come from?

If the answer is essentially, “ChatGPT estimated them,” then you don’t have a budget.

You have an assumption.

And assumptions are useful at the beginning of a planning process. They are not a substitute for the planning process itself.

When you start getting actual quotations, confirming quantities, negotiating supplier rates, calculating logistics and accounting for the smaller costs that are easy to overlook, that KSh 841,000 can very quickly become KSh 1.2 million.

The problem isn’t that AI got the answer wrong.

The problem is that we treated an estimate as fact.

The missing mathematics

One of the things I find particularly concerning is how often we are making commercial decisions without doing the mathematics behind them.

Someone says:

“Let’s charge KSh 5,000 per ticket.”

Why?

“It sounds reasonable.”

Or:

“Other events are charging around that.”

But that isn’t pricing.

At the very least, you should know:

• What is the total cost of delivering the event?
• How many people realistically will attend?
• What percentage of tickets can you reasonably expect to sell?
• What are the fixed costs?
• What are the variable costs?
• Are taxes included?
• Are transaction charges included?
• Are marketing and communication costs included?
• What happens if supplier costs increase?
• What is the break-even point?
• What margin are you targeting?
• What happens if you sell 70% rather than 100% of the tickets?

Only then can you begin to answer whether KSh 5,000 is the right ticket price.

You don’t need artificial intelligence to tell you that.

You need a spreadsheet.

You need quotations.

You need assumptions.

You need someone to challenge those assumptions.

And you need to do the maths.

The same problem exists beyond events

This isn’t just about event management.

It applies to almost every business decision.

A company wants to launch a product.

AI can help develop the launch strategy.

But AI doesn’t replace market research.

A business wants to run an advertising campaign.

AI can suggest a media plan.

But it doesn’t replace knowing what your media actually costs, who your audience is, how much inventory is available or what your historical conversion rates look like.

A company wants to hire ten people.

AI can help develop job descriptions and estimate salary ranges.

But the actual organisation still needs to establish its salary structure, statutory costs, recruitment costs, onboarding requirements and total employment cost.

Someone wants to import a vehicle.

AI can provide a useful starting point for understanding the process.

But the actual decision needs to be based on the vehicle’s purchase price, freight, taxes, clearing, registration, insurance, local transport and all the other costs that apply to that particular transaction.

The principle is the same everywhere:

Use AI to help you think. Use actual data to make the decision.

Process still matters

There is something else we risk losing in our enthusiasm for AI: process.

Good organisations have processes for a reason.

You gather requirements.

You establish scope.

You identify the resources required.

You obtain quotations.

You compare suppliers.

You calculate costs.

You identify risks.

You establish assumptions.

You develop a budget.

You test the budget against different scenarios.

You determine your break-even point.

Then you make a decision.

AI can assist with almost every one of those steps.

But it shouldn’t allow you to skip them.

If anything, AI should make the process better and faster, not make the process disappear.

The danger of the confident answer

Perhaps the most dangerous thing about AI is not that it can be wrong.

It is that it can be wrong while sounding incredibly convincing.

A poorly researched answer from a human may look obviously incomplete.

An AI-generated answer can arrive beautifully formatted, professionally written and accompanied by precise-looking figures.

That creates a false sense of certainty.

And this is where professional judgement becomes even more important.

Before accepting an AI-generated figure, ask:

What is the source?

What assumptions were made?

Does this reflect my market?

Does this reflect my project?

Have I independently verified it?

What information could change the figure?

Those five questions can save a business a lot of money.

AI should challenge our thinking, not replace it

I am a strong believer in using technology to improve how we work.

There is no reason to reject AI.

In fact, I think organisations that learn to use AI properly will have a significant advantage.

But proper use means understanding what the tool is good at.

Ask AI to:

• help you structure a project;
• identify potential cost categories;
• develop scenarios;
• challenge your assumptions;
• analyse data you provide;
• identify gaps in your plan;
• generate questions you should be asking;
• help you communicate the final plan.

Then bring in the real-world information.

Get the quotations.

Talk to the suppliers.

Check the rates.

Confirm the quantities.

Review the contracts.

Understand the market.

Do the mathematics.

And then make the decision.

The human responsibility remains

There is a temptation to say:

“ChatGPT said…”

But ChatGPT isn’t responsible for the decision.

You are.

Your organisation is.

Your client is.

Your investors are.

Your customers are.

If an AI-generated budget is wrong and your event makes a loss, you cannot tell your supplier, your client or your shareholders that ChatGPT gave you the figure.

The responsibility for validating information and making the final decision remains with the person using the information.

That is why the more powerful AI becomes, the more important critical thinking, professional judgement and proper processes become.

Don’t let convenience become negligence

There is nothing wrong with wanting to work faster.

There is nothing wrong with using AI to save time.

There is nothing wrong with asking ChatGPT to give you a starting point.

The problem starts when convenience replaces diligence.

A five-minute AI-generated budget may save you an hour today.

But failing to validate it could cost you hundreds of thousands of shillings tomorrow.

The objective shouldn’t be to eliminate the process.

The objective should be to use AI to make the process better.

Because ultimately, good decisions still require good information.

And good information comes from accurate data, proper research, actual figures and a willingness to do the work.

AI can help you get there.

But it cannot do the due process for you.