I have been delivering artificial intelligence training to companies and, increasingly, I find myself thinking about what happens afterwards. During the session, someone discovers they can complete in twenty minutes a task that used to take two hours. They try it, confirm the result and begin to imagine where else they might apply it. But the training ends, people return to their desks and, the following morning, the company continues to operate exactly as it did before.
Consider a transport company in Maputo. The fleet manager learns to analyse vehicle data with AI and, within minutes, has a report that previously consumed a good part of his morning. He then prints it, takes it for signature, sends it to the finance department and waits for someone to manually re-enter, into another system, the figures that were already in the first. The analysis became faster; the report continues to make the same journey.
It is along that journey that a share of the gains is lost. The individual has changed the way they work, but remains dependent on procedures designed for a time when everything took longer. They can clear their workload before lunch and spend the afternoon waiting for an approval. Seen from their desk, there has been an enormous step forward. For the client awaiting a response, nothing may have changed yet.
In a small company, this gap tends to be shorter. Someone who divides their day between attending to clients, preparing proposals and handling invoices generally knows what to do with a free hour. There is always something postponed, waiting for exactly that. But even there, the gain can stall at the owner's office door if every decision requires their sign-off and they still have no time to decide.
In larger organisations, change touches a structure built over years. Functions were created to gather information, verify documents and prepare work for the next person in the chain. Teams and departments grew around those needs. When certain tasks begin to take minutes rather than hours, it is worth revisiting those circuits. Some controls protect the business and should remain. Others exist for reasons that nobody can now clearly explain.
We have seen something similar before. For years, the director wrote by hand and the typist transcribed the text. With the arrival of the computer, many directors began writing their own documents, and that division of labour gradually lost its purpose. AI forces us once again to examine those boundaries — this time with the capacity to alter several functions simultaneously.
The conversation becomes more sensitive when we move from tasks to people. On realising that part of a job can be automated, it is tempting to count the hours saved and think only about cost. Yet the spreadsheet used for that calculation rarely has a column for what an employee has learned over fifteen years with the company.
That employee knows why a particular procedure was created, which client requires a different approach and where the system says one thing but operations demand another. They recognise an unlikely number and recall a problem that arose before, even though nobody recorded how it was resolved. They may have been hired to prepare reports. Over time, they have come to know far more than how to prepare reports.
If AI frees up 40% of their time, it is worth examining that accumulated knowledge alongside the work that remains undone. There may finally be capacity to attend to clients who only receive attention when they complain, to resolve matters that drift from desk to desk, or to verify whether the new automations are performing as expected. Many companies have lived for years with needs to which no one has been able to respond.
That transition requires preparation. It is not enough to announce that, from Monday, someone will have a more strategic role. It is necessary to understand what they already know how to do, what they will need to learn and who will support them through the change. A capable data analyst may need guidance in managing client relationships or taking on decision-making responsibilities. The next training programme may not be about AI at all.
Some companies are already actively working on this reorganisation, reviewing procedures and preparing their teams for broader responsibilities. And there are those who support that process, connecting knowledge of the tools to an understanding of how work actually gets done inside each organisation. While some businesses are still debating whether to adopt AI, others are already addressing the day after.
As I deliver these training sessions, it becomes clearer to me that part of the outcome depends on those who lead the company. I can teach someone to prepare an analysis or to work more effectively with a document. I cannot, from the training room, alter the circuit that requires that document to pass through five people before reaching the decision-maker. That responsibility belongs to management and is not resolved by purchasing software licences.
So when we switch off the projector, a question remains — one that deserves to reach those who make such decisions: if tomorrow everyone in this company is proficient with these tools, which parts of our organisation no longer make sense, or need to be redesigned?
The answer will take work. That is the work of the day after AI.
Source: Diário Económico
Original article: https://www.diarioeconomico.co.mz/?p=547949