Businesses are terrible at making decisions.
Everything about the way we build organisations is designed for coordination. Hierarchy, committees, sign-off chains, governance. All of it exists to slow things down, spread the risk and keep any one individual clear of the consequences. It's why the most reliable sentence in corporate life is let's take this to a meeting. That line has never once accelerated a resolution. It diffuses ownership, and that is precisely the job it's doing. I've sat in enough of those rooms to recognise the move, and I've made it myself.
For most of the last two hundred years it didn't cost us much, because the thing we actually competed on sat upstream of the decision. Information was the moat. If you understood your customers' buying patterns, the market, your suppliers' capacity to take on new work better than the business down the road, you were in the strongest position in the room. Being slow to decide was survivable when you were the only one who knew.
That advantage has disappeared. Any organisation can now synthesise vast data sets and surface something useful in seconds. Information becomes a commodity, like everything else, and what's left to compete on sits either side of it: who is asking the right question, and who can do something with the answer.
Both of those are decisions. Which leaves us competing on precisely the capability we designed our organisations to avoid exercising.
The second thing in the way is emotion. Bid decisions are the cleanest example I know. A large tender lands. The account exec has history with that client and wants it badly. Finance and the delivery team can see it isn't winnable, and nobody can articulate why. So the business bids. It spends the time, the energy and the cost on something it already knows it will lose, because the strength of feeling outweighed the evidence on the table.
That is the real case for putting a machine near the decision. It's tireless and emotionally uninvested. Feed it a tender and it will come back with a weighting, your win points, your weak areas, and the things you would need to fix to make it viable.
I'm not that interested in whether AI can make our decisions. In a narrow sense it already can, faster and with less bias than the committee, on whatever data we've given it. The question I keep coming back to is whether we're prepared to own what comes out.
I'd put my money on governance ahead of models. The organisations that get ahead will be the ones that can say who owns the call, on what evidence, and what happens when it goes wrong. They'll also need the cultural confidence to act on a recommendation faster than their competitors, which is a harder combination than it sounds.
None of that happens by default. The default is that we carry on coordinating, carry on taking it to a meeting, and now we have a machine to point at when it goes wrong. Deciding who owns the call, and on what evidence, is the deliberate choice. That work is vital, and it can't be handed to the thing making the recommendation.
I got into all of this with Simon Cuthbert of Arthurian Labs on this week's episode of The Next Thing Now.