Asher's Blog

Across many African companies, important operations are held together by spreadsheets maintained by a small number of experienced managers. Payroll schedules, sales forecasts, stock records, compliance deadlines, reconciliations and staff reports live inside files that move through email and WhatsApp. The manager knows which tab is current, which formula must not be touched and which number needs to be adjusted before the report reaches leadership.

The spreadsheet is not the problem by itself. It became popular because it is flexible, familiar and available. The problem is that the company’s operating knowledge often remains trapped inside a document that cannot explain itself, enforce consistent process or act on what it knows.

AI will not kill spreadsheets completely. It will kill the spreadsheet’s role as the invisible operating system of the company.

The spreadsheet compensates for missing software

Businesses use spreadsheets where formal systems are too expensive, too rigid or poorly adapted to local work. A capable manager creates formulas, colour codes and manual checks that allow the organisation to keep moving.

This ingenuity deserves respect. The spreadsheet often represents years of adaptation to reality. Replacing it without understanding the work can make operations worse.

The opportunity is to capture the logic, decisions and exceptions the spreadsheet contains, then build an intelligent workflow around them. The new system should preserve flexibility while adding controls, collaboration, history and automation.

Middle managers spend too much time moving information

Many managers are employed partly as human integration layers. They collect reports from several teams, correct inconsistencies, update the spreadsheet, prepare a summary for senior leadership and return to the teams for explanations.

This work is important because the organisation lacks another way to coordinate itself, but it uses experienced people for repetitive information movement. Intelligent systems can collect data directly, identify missing inputs, reconcile differences and generate a current view without waiting for the monthly reporting cycle.

The manager’s value then shifts from maintaining the spreadsheet to interpreting the business. Why is one region declining? Which cost signals an operational weakness? What decision should leadership make now rather than after month-end?

AI should elevate the manager from report producer to decision owner.

The back office becomes conversational

Traditional software requires the user to know where information lives and how the system is organised. An intelligent back office allows people to ask questions in ordinary language: which employees have unresolved compliance records, why did payroll cost rise, which invoices are approaching their due dates, or what changed in customer acquisition this month?

The system can retrieve evidence, explain calculations and propose next actions. This is particularly valuable to growing African businesses that cannot afford a specialist analyst for every function.

However, conversational access must not become confident improvisation. Every important answer should connect to authoritative data, identify uncertainty and preserve a trace of how the conclusion was produced. A fluent answer without evidence is more dangerous than an obvious spreadsheet error.

Data entry should become the exception

Back offices spend enormous time re-entering information that already exists somewhere else. Employee records are copied into payroll, payment outcomes are copied into reconciliation files and sales information is copied into management reports.

AI and better integrations can extract, classify and match this information, leaving people to review exceptions. This changes the workflow from “enter everything correctly” to “investigate what the system cannot reconcile confidently.”

The design must account for local realities. Documents may have inconsistent formats, names may appear differently across systems and third-party infrastructure may return incomplete information. The intelligent system should expose uncertainty rather than silently force every record into a false match.

Management must not disappear with administration

The death of the manager’s spreadsheet does not require the death of the manager. Organisations still need people to set priorities, resolve trade-offs, coach employees and accept responsibility for outcomes.

Some middle-management roles exist mainly because information moves slowly through the organisation. Those roles will change or disappear. Others will become more important because the person can now spend less time assembling facts and more time exercising judgment.

Companies should prepare people for that transition. Teaching a manager to use an AI tool is not enough. The manager must learn to ask better questions, verify evidence, understand economics and make decisions rather than merely forwarding reports.

The spreadsheet must be understood before it is replaced

When transforming a back-office process, sit with the people who currently operate it. Watch what they do, including the adjustments they make outside the documented workflow. Ask which errors they fear and which checks give them confidence.

The formulas are only one part of the system. The manager may know that one supplier submits data late, one branch uses a different convention and one unusual transaction requires separate approval. If these details are ignored, automation will remove the person who compensated for them without removing the underlying exceptions.

The manual spreadsheet becomes the specification for the new product, but not every historical step deserves to survive. Some exist only because the old tools were limited. The redesign should preserve necessary judgment and eliminate inherited friction.

Africa can build intelligent back offices directly

Many African small and medium-sized businesses have not completed a long transition through generations of enterprise software. This appears like a disadvantage, but it creates room to adopt systems designed around intelligent assistance from the beginning.

A business should not need to purchase several complex tools, hire administrators to operate them and then add AI later. It can use one coherent system that captures transactions, maintains records, manages obligations and explains what requires attention.

This is the leapfrog at company level: moving from manual spreadsheets to intelligent workflows without reproducing every layer of expensive legacy software.

The product must still be affordable, secure and adapted to the market. It must work with the infrastructure businesses actually use and make migration from existing files simple. Telling customers to abandon familiar tools without preserving control will slow adoption.

The future belongs to the manager who can decide

The middle manager’s spreadsheet will not vanish in one dramatic moment. Intelligent functions will appear around it: automatic classification, anomaly detection, conversational analysis, reconciliation and workflow execution. Eventually, the spreadsheet will remain only where its flexibility is still genuinely useful.

The deeper change is organisational. Information will travel faster, routine reporting will require fewer people and leaders will expect decisions closer to real time. Managers whose value came from controlling access to information will struggle. Managers who can interpret, coordinate and take responsibility will become more valuable.

AI will not merely improve the spreadsheet. It will expose how much management has been organised around maintaining it. Once the machine can move the information, the human must show that they can move the organisation.