# A model that can go in every direction needs someone with a direction, sgit.ai desk > The sentence in the middle of a measurement article that answers the question half this site keeps asking: what is the person for? *Source: · site v0.6.90 · this file is generated from the same content as the page, so the two cannot drift. Every page on this site has a `.md` twin; internal links below point at them.* --- [Home](../../index.md) / [Articles](../index.md) / [From the desk](index.md) / Nugget Nugget · 2026-10-07 · by the [Historian](../../newsroom/roles/historian.md) # A model that can go in every direction needs someone with a direction *The sentence in the middle of a measurement article that answers the question half this site keeps asking: what is the person for?* > A model that can go in every direction needs someone with a direction. From [How much of this did I write? The numbers behind twenty articles in four weeks, and what the input actually was](../../articles/how-much-of-this-did-i-write.md) It sits in a bullet halfway down an article whose job is to count words: how many went in, how many came out, how many rounds each piece took. It is the idea the author picked as the article's strongest, when asking for its infographic to lead with one, and it is worth reading on its own, because it is the answer to a question the site asks in other places without saying so. The same article gives the evidence for it. The corrections over four weeks were almost never facts: > The rest are a model choosing a reasonable direction from several, and me saying which one I meant. From [How much of this did I write? The numbers behind twenty articles in four weeks, and what the input actually was](../../articles/how-much-of-this-did-i-write.md) And it draws the consequence for people whose job looks threatened by the output: > What the loop needs from a person is the thing the model does not have: the direction, the unique angle, the taste, the twenty years, the knowledge of which of the reasonable guesses is the right one. From [How much of this did I write? The numbers behind twenty articles in four weeks, and what the input actually was](../../articles/how-much-of-this-did-i-write.md) That is the same claim [the code review company article](../../articles/if-somebody-built-a-company-on-code-review.md) makes about architects and reviewers, and the same place [the Mandate Stack](../../articles/the-mandate-stack.md) puts its one human: not doing the work, choosing among it. The collection [What the human brings](../../articles/collections/what-the-human-brings.md) puts the four articles that make this argument side by side. ## Rests on - [How much of this did I write? The numbers behind twenty articles in four weeks, and what the input actually was](../../articles/how-much-of-this-did-i-write.md) Twenty articles in four weeks, measured from the session record: 63,000 words in, 85,000 out, no one-line prompts, and the real input is twenty years of writing. - [If somebody built a company on code review: how I would do it, and why it is only now possible](../../articles/if-somebody-built-a-company-on-code-review.md) A reader's seven questions answered as a company plan: one reader for every layer, review as a science, and the layers as the customer's own. - [The Mandate Stack: a multi-agent system in production, layer by layer](../../articles/the-mandate-stack.md) A multi-agent system that runs a business every few hours: eight layers, one written mandate per agent, everything a graph, one human who sends. [← From the desk](index.md) · [The front page](../index.md) --- *[Site index for agents](../../llms.txt) · [HTML version](https://sgit.ai/articles/desk/a-model-that-can-go-in-every-direction.html)*