Ship or Kill

The phrase going round is service as software. It describes work that used to arrive as advice and now arrives as something that runs: available when the client needs it, consistent between the people delivering it, better in this version than it was in the last one.

Software companies have already moved into work that used to be billable hours. Contract review, bookkeeping, candidate screening, first-line support. That direction is settled. What is open is which firms end up on the delivering side of it.

Nobody has finished this transition. Any firm claiming to have arrived is describing an ambition. The work behind it is still done by people, by hand, at consulting rates. Ours is.

The asset is a structured, current representation of how the firm works: the processes it has approved, the policies it has decided, the templates people actually use, and the judgement calls it has already made and does not want to make twice. Written in a form a machine can read and a person can correct.

Two firms. Same size, same clients, same year. The first writes its decisions into a shared record and keeps that record current. The second keeps them in conversation. By December the first firm is producing the same quality at a lower cost to produce it, because most of the thinking was done once and is still in the building. The second firm pays full price for it every time.

Every decision written down narrows the next one. That is what compounding means here, and it is the whole argument.

The firms most exposed are the ones whose economics reward keeping the method in people's heads. An hour billed is an hour in which the knowledge stayed exactly where it was. That arrangement held while the client's alternative was another firm doing the same thing by the same means. The client's alternative is what changed.

The AI numbers say it in a different register. 88% of organisations now use AI regularly in at least one part of the business, on McKinsey's 2025 reading, and fewer than 4 in 10 can point to any effect on profit. Cisco found that the organisations getting real results track the impact of what they build, 95% of them, 3 times the rate of everybody else. The difference sits in whether anybody wrote down what good looks like and then kept it current.

The World Economic Forum's 2025 Future of Jobs report puts 39% of core workforce skills changing by 2030. A firm that cannot state how it works today has nothing to hand its people when that lands.

Who wins the shift: firms that treat their operating knowledge as an asset with an owner and a maintenance schedule, the same way they treat a client list.

Who gets stranded: firms whose knowledge is real, valuable and undocumented, which from the outside is indistinguishable from not having it.

The firm that gets this right is recognisable from the outside.

A new hire starts from the firm's approved way of doing the work instead of reconstructing it from whoever is free that afternoon. Every funded initiative has one name against it, a metric that person agreed to, and a date it gets reviewed. Work that used to wait for the one person who knew how now moves without them. The record of how the firm operates is current, because keeping it current is somebody's job on a fixed rhythm rather than a project that someone finishes. Quality stops depending on which partner picked up the file.

None of that needs a new model or a new vendor. It needs the firm to write down how it works, and then to keep the writing honest, which is the harder half.

The discipline underneath all of it fits in a paragraph. Something gets proposed. One person owns it. It carries a metric and a date. On that date it either goes live in the business or it gets killed in writing, with the reason recorded and the capacity moved somewhere else.

Kill is the half that gets skipped, and it is the cheaper half. A parked pilot costs attention every week it stays parked and returns nothing. Firms that are good at this kill early, kill in writing, and stop paying for a decision they have already made.

The record of those decisions is the thing that compounds. Every kill narrows the next choice. Every workflow that goes live is one the firm stops arguing about. That record, kept current, is the representation of how the firm works that the shift rewards. The two halves are the same job.

Two things are true about the business publishing this page.

The first is the argument above. Expert work is becoming software-delivered, and firms that own a current, structured account of how they work will compound while the rest pay full price every time.

The second is that every install this business has run was run by hand, by one person, and the recurring work above it is priced as consulting. 200 seats of handwritten monthly notes is 50 hours a week. That is arithmetic we have already run on ourselves, and it is why the client's record gets built as something a system can read rather than something anybody retypes.

Stating that costs less than pricing around it.