Imag­ine a shop­keep­er in Kisumu stock­ing up for a busy week­end, her shelves full and her mon­ey com­mit­ted, when the rain arrives and qui­et­ly shifts what her cus­tomers want. By the time she notices, some of that stock has become a loss. What if she had known two days earlier?

That ques­tion fol­lowed me home from the exec­u­tive forum PwC Kenya orga­nized on Thurs­day, Sep­tem­ber 24. One demon­stra­tion com­bined weath­er, sales, and stock data, saw that rain would soft­en demand in Kisumu, and pro­posed mov­ing crates to Naku­ru, where shelves were near­ly bare. It took less than a minute, then stopped because the val­ue crossed a lim­it its own­ers had set and asked a human to approve. I smiled because I have known men who move far more than that with­out ask­ing anyone.

Beyond my dai­ly use of these tools, that moment clar­i­fied every­thing. AI can detect a pat­tern ear­ly enough for us to act, weigh what it sees, explain itself, and offer a choice. What it can­not do is under­stand the cir­cum­stances of a fam­i­ly in Kisumu or shoul­der the blame when its advice is wrong. That respon­si­bil­i­ty remains ours.

A sec­ond demon­stra­tion involved read­ing sup­pli­er invoic­es of every shape and check­ing them against the tax reg­is­ters, a task any­one who has lost a Sat­ur­day to paper­work will appre­ci­ate. One find­ing stayed with me. An invoice with flaw­less arith­metic was still wrong because tax had been charged on an exempt line. 864 shillings that no rule would have caught were found because some­one went back and looked again.

Now look beyond the shop, where a farmer is decid­ing when to har­vest, a clin­ic is try­ing not to run out of med­i­cine, a coun­ty should be clear­ing its drains, and fore­cast­ers expect these short rains to arrive ear­ly, at about 150 per­cent of aver­age. We could know which bore­holes have failed, which drains clog first, and which of the trees we plant­ed are still alive. The ques­tion is the same: what do we need to know soon­er, and what would we do dif­fer­ent­ly if we knew it?

This is where lead­ers must stop admir­ing tools and start nam­ing prob­lems. PwC’s sur­vey of 4,454 chief exec­u­tives found that only 12 per­cent saw both high­er rev­enue and low­er costs, and 56 per­cent saw no real ben­e­fit. Mean­while, 64 per­cent of African work­ers say they already use it at work. Our peo­ple are mov­ing, while our insti­tu­tions are still clear­ing their throats. If your team is wait­ing for a pol­i­cy nobody is writ­ing, that is not cau­tion; it is miss­ing motion at what I call the speed of life.

When asked what holds them back, near­ly half of us who attend­ed the PWC ses­sion at the Cap­i­tal Club point­ed not to the machines but to their own records. That should sober us, for it is our old­est weak­ness. We count what we plant, not what sur­vives; we announce and rarely return to ver­i­fy. A sys­tem that sens­es but nev­er ver­i­fies is only a faster pho­to­graph. No com­pa­ny or coun­try can be wis­er than what it has hon­est­ly recorded.

Keep ask­ing the uncom­fort­able ques­tions, espe­cial­ly when liveli­hoods, health, or rights are at stake. Could pri­vate infor­ma­tion be leaked? Might the advice be unfair? What does it cost in mon­ey, ener­gy, and water? Who is account­able if a deci­sion harms someone?

My friends across Kenya and Africa, let’s not send our peo­ple on a course and call it a strat­e­gy. Let’s sit with those who live the prob­lem, ask what they need to know soon­er, test a small change, then return to see whether it helped. Sys­tem trans­for­ma­tion starts right there.

We now have e‑agents mon­i­tor­ing rain­fall at a ware­house, and the same fore­cast should prompt drainage crews, fod­der, seedlings, and improved effi­cien­cy across our insti­tu­tions. The rain is com­ing, so let us be the gen­er­a­tion that saw it ear­ly and act­ed in time. Think Green, Act Green.