Buy, build, borrow, bot

Every leader now faces the same recurring question, only faster than before: a capability is needed — a skill, a function, a piece of work — and it is not on the team today. The old reflex was to post a job and hire. But hiring is now just one of four routes, and choosing the wrong one is expensive in time you do not have. You can buy the capability, build it, borrow it, or hand it to a bot.

This is not semantics. Each route carries a different cost, speed, risk and half-life, and the gap between them is widening as AI makes “bot” a serious answer to questions it never used to touch. Treating every gap as a hiring problem — or, lately, automating every gap because you can — is how organisations end up over-staffed in the wrong places and dangerously thin in the right ones.

And running underneath all four is a question leaders too rarely ask out loud: which capabilities are so core to our judgement and trust that they must stay human, whatever the tools can do?

Start by naming them plainly. Buy is hiring the capability in — fast access to proven skill, but expensive, scarce, and a bet on one person staying. Build is growing it from the people you have — slower and uncertain, but it compounds, deepens loyalty, and keeps knowledge in the house. Borrow is renting it — contractors, agencies, fractional experts — ideal for a spike or a specialism you need now but not forever. And bot is automating it — handing a defined, repeatable task to software or an AI agent.

The error is rarely choosing one of the four. It is choosing reflexively — defaulting to “buy” because that is what HR knows how to do, or to “bot” because the budget loves it — without asking which route the capability actually wants. A rule of thumb that has aged well: build what is core and enduring, buy what is core and urgent, borrow what is peripheral and temporary, and bot what is repeatable and rule-bound. The skill is matching the route to the work.

The error is rarely choosing one of the four. It is choosing reflexively, without asking which route the work wants.

Two questions settle most cases. First: how core is this capability to our strategy and judgement? Second: how repeatable and stable is the work — or is it ambiguous and changing? Plot a gap against those two axes and the right route usually announces itself. Core and ambiguous work is where you invest in people; peripheral and stable work is where automation and contractors earn their keep.

The matrix is not a machine that decides for you, but a discipline that stops you defaulting. It forces the two questions to be asked before the cheque is written — and it makes visible the most dangerous square of all: core, high-judgement work quietly handed to a bot because it was technically possible, not because it was wise.

Here is the variable that wrecks tidy resourcing plans: skills now have a half-life, and it is shrinking. The WEF expects 39% of core skills to change by 2030, with 59% of workers needing training in that window. A capability you buy at a premium today may be commodity — or obsolete — in three years. This is precisely why the four routes are not interchangeable: each ages differently.

Buy and borrow give you today’s skill at today’s price, but you have rented a depreciating asset; the half-life clock starts the moment they arrive. Build is slower, but because you are growing a learning capacity rather than buying a fixed skill, it ages better — people you develop can be developed again. Bot looks permanent but quietly rots too: an automation is only as current as the process it encodes, and processes change. The lesson is to resource for the half-life, not the headline — and to keep building the one thing that does not expire, which is the capacity to learn.

Because “bot” is now cheap and capable, the most important resourcing decision is increasingly a decision not to automate. Some capabilities can be done by a machine and still should not be — not for sentiment, but because automating them quietly erodes the trust, judgement or accountability the organisation runs on. The conversation that ends someone’s employment. The final sign-off on a safety-critical call. The relationship a key client believes is with a person. The ethical judgement that has to be owned by a name, not a model.

Draw that line deliberately, while you still can. Once a capability has been automated and the humans who held it have moved on, it is extraordinarily hard to rebuild — you have lost not just the task but the tacit judgement that surrounded it. The strongest organisations of the next decade will be defined as much by what they refused to hand to a bot as by what they automated. Efficiency is easy to add later; judgement, once lost, is not.

The most important resourcing decision is increasingly the decision not to automate.

The aim is not a perfect plan but a deliberate one — every gap matched to a route, on purpose rather than by reflex.

Do this and resourcing stops being a reactive scramble and becomes what it should be: the deliberate shaping of where, and how, your organisation chooses to be human.

Buy, build, borrow or bot is a strategy, not an HR errand. Choose on purpose — especially when you choose to stay human.