Business Software Was Built to Count

Why Legacy Business Software Specializes in Counting

Counting is often at the heart of the job in business management.

How many new leads do we have?

How many calls did I make today?

How much should I spend on lunch?

We count in units, inches, pounds, and dollars.

Spreadsheets have always been optimized for this. We enter a SUM formula, highlight a column, and see the total.

I remember the commercial rise of Microsoft Excel in the early 1990s.

This was when legacy software, as we know it, was born.

Legacy software is built for counting and adding. It is built for the thumb managers, or in other words, business professionals who manage through columns, totals, and reports.

A large portion of the population went to college and learned how to become a counter.

But business is more than counting.

It is the product.

It is the brand.

It is the relationships between people, assets, and geography.

So why does legacy software specialize in counting?

And why do more small businesses not use legacy platforms to manage their businesses?

Great Plains, Microsoft Dynamics GP, and the Top Down Model

The history of business software tells the story of macroeconomic trends.

In the 1990s, the internet was still in its infancy. If you knew how to write software, you were highly sought after.

Applications such as Excel were widely distributed.

Specialty software was not.

The specialty platforms that quietly began to appear were systems such as Great Plains, also known as Microsoft Dynamics GP.

They were accounting machines, and much of the demand came from publicly traded companies.

For that model to succeed, legacy software was built with a top down approach.

At the top was the CEO.

Then the CFO.

Then the board.

Then the shareholders.

If your name was Larry the shoeshine guy, you never received an invitation to the party.

This is exactly why Main Street and Wall Street can seem so disconnected from one another.

The emphasis has traditionally been on macroeconomics, often at the expense of microeconomics.

Larry the shoeshiner does not care about the depreciation of his chairs.

His goal is to book, operate, account, and provide five star service.

CRM Added Attributes, Not Understanding

Around 2010, enterprises began to grow bored with simply counting transactional currencies.

They wanted a new breed of software.

This was customer relationship management, or CRM.

What had once been business software for bean counters began to evolve into customer service teams entering poorly written text into open input fields.

At the exact same time, salespeople in the field were using early iPhones to enter field logs through desktop mode websites on tiny displays.

BlackBerries could not do this in the same way.

By today’s standards, these systems were primitive.

The C suite had realized that, in order to squeeze out larger margins, tracking attribute information was important.

Who is the customer?

What did they ask for?

When were they contacted?

What product were they interested in?

What happened during the last conversation?

The problem began to surface.

The information existed, but it was scattered across systems, departments, fields, and databases.

Then, very quickly, came the solution:

SQL Server Integration Services, or SSIS.

The Consultants Never Left

Entire teams and consulting firms became dedicated to building integration services.

The pitch to C suite executives was simple:

We will eventually set you up with an all in one system.

The consultants never left.

The story was far bigger than the reality.

During the 2010s, it sometimes seemed as though half of enterprise IT spending went toward consultants trying to make disconnected systems behave like one system.

Fast forward to today.

The new story is the same trick.

AI is the new integration.

AI spending is out of control.

If only the problem could be solved with something as small as a sudo nano .py patch.

That phrase is important.

The sudo nano .py patch is my own term for a small, direct change that solves a real problem without requiring another enormous platform, consulting engagement, or integration project.

Open a file.

Make the change.

Test it.

Move forward.

AI is now being sold as the promise that will bridge the gap between systems held together by spaghetti string integrations and systems that operate as true all in one business operating systems.

But that promise will never be realized by sprinkling AI on top of legacy software.

Modern business software, or MBS, not to be confused with the Saudi crown prince, has not emerged yet.

Some companies have appeared close. Then an incorrect prompt or model response exposed how much of the operating system was actually being controlled by a fragile bot.

A business should not be told that it has dependable operating infrastructure when the infrastructure is really an unpredictable chain of prompts.

That should not be allowed.

A modern business system should capture the entire picture of a business.

Rather than being built on legacy software architecture, it should be composed of a minimal number of database tables.

AI Should Not Be Sprinkled on Top

Artificial intelligence should not be sprinkled on top of systems that already exist.

The cost of moving and rebuilding software has fallen dramatically. Language models can help interpret old systems, rewrite logic, migrate records, and build new interfaces.

That means new architecture is not only possible.

It is necessary.

Historically, database systems were built with many tables and relationships because humans had to review and manage the data.

Every concept received a table.

Every relationship received another table.

Every new requirement introduced another layer.

New frameworks should move toward what I call a one GL methodology.

One central record of truth.

Many columns.

Fewer tables.

Language models can navigate database tables containing thousands of columns. They can interpret what those attributes mean, locate the relevant information, and help manage the relationships between them.

Enough with the architecture.

A modern business system keeps track of two things:

Money and attributes.

It does not need a million connectors. A connector should exist when the system must communicate with something external, such as sending correspondence through an email provider.

Payment gateways should be embedded directly into the customer experience.

Contracts and document management should be attached to the same GUID business record.

Images should be captured, identified by a language model, and consumed as structured details.

Retrieval should happen through RAG, or retrieval augmented generation, so that a business can ask questions about its own records and receive answers grounded in its own information.

The system should understand the business as a whole.

Not as a collection of tabs.

Not as a series of integrations.

Not as ten different products pretending to be one.

The All in One That Was Always Promised

Every day, entrepreneurs start businesses around the world.

Why?

They have a skill that the world deserves to discover.

A product to sell.

A service to provide.

In a capitalist world, products and services can become worse over time when control becomes concentrated and competition disappears.

Trickle up economics creates a different possibility.

Competition on a level playing field proves who the winner is.

A modern business system supports trickle up economics.

It gives a small business access to the kinds of tools, information, and automation that were once reserved for publicly traded companies.

It does not begin with the CEO, the CFO, the board, and the shareholders.

It begins with Larry the shoeshiner.

What does Larry need to do today?

Who booked?

Who paid?

What service was provided?

What reminder must be sent?

What does the customer need next?

A modern business system is the all in one that was always promised.

It is attributes.

It is accounting.

It is reminders.

It is notifications.

It is payments.

It is the brand.

It is also flying cars, street food, rent a chef, and golf cart rentals.

Business software was built to count.

Modern business software must learn to understand the business being counted.