The 24-month window

Every operational technology has had a window: a period when adopting it early conferred a lead that later adopters paid heavily to close. ERP had one. Cloud had one. Agentization has one now, and it is shorter.

The pattern is consistent. In the first phase, a technology is treated as an experiment — pilots, committees, cautious budgets. In the second, the early movers publish results and the market re-prices the risk: suddenly the danger is not adopting too early but too late. In the third, the technology is table stakes and confers no advantage at all. The window is the second phase. It is where leads are built.

Why is this window shorter? Because agents, unlike ERP or cloud, improve the moment they are deployed. An ERP system installed in 2003 did roughly what it did in 2008. An agent system deployed this year learns the enterprise it runs in — its vocabulary, its approval chains, its exceptions — and compounds that knowledge monthly. The gap between a firm eighteen months into production and a firm starting cold is not eighteen months of licence fees. It is eighteen months of accumulated adaptation that the follower cannot buy.

There is a second compounding asset: people. The scarce skill of the next decade is directing agents — knowing what to delegate, where to set boundaries, when to intervene. That skill is learned in production, on live workflows, under real stakes. First movers are training it now. Their followers will recruit it later, at a premium, from the first movers.

Our estimate of twenty-four months is deliberately conservative. It assumes regulators move slowly and incumbents move slower. In sectors where one aggressive competitor forces the pace — insurance claims, trade finance, logistics — the window is closing faster.

The strategic question is therefore not whether the estimate is right to the month. It is whether your enterprise is spending the window deciding, or spending it learning.

Start a conversation →