The difference is repeatability. Using it means opening a chat when something feels tedious and typing whatever comes to mind. A workflow means the same task, entering the same way, producing output that already sounds like you, without you reconstructing the context each time.
The gap between those two is one document and about four weeks of habit.
The asset everything depends on
If you build nothing else from this, build this.
Keep one document containing everything a capable assistant would need to know about your business before helping with anything. Not prompts. Context.
Specifically:
- The sentence describing who you serve and what changes for them
- Three examples of your actual writing, chosen because they sound like you
- Your services, what each includes, and what each costs
- Who your customers are, described the way they would describe themselves
- The words customers use for their problems, taken from real emails rather than paraphrased
- The five objections you hear most, and how you answer each
- What you explicitly do not do, and who you are not a fit for
- Your standard terms, timelines, and process
Two pages is plenty. This is the material referred to elsewhere as the object, and it is the difference between output that sounds generic and output that sounds like your business.
Everybody has access to the same model. Almost nobody has assembled the specific information that makes it useful for their business. That document is the entire advantage, and it takes an afternoon.
Three loops, not one
A workflow is not one habit. It is three, running at different speeds.
The fast loop, several times a day. Short tasks with immediate judgment. Rewriting an email you already drafted. Turning meeting notes into an action list. Explaining something back to yourself to check you understand it. These take under two minutes and you evaluate the output instantly.
The weekly loop, batched. Work you deliberately collect rather than handle as it arrives. Drafting the week of follow ups at once. Turning one piece of thinking into several formats. Reviewing what came in and deciding what matters. Batching is what makes this efficient, because you paste the context once and work through six things.
The project loop, occasional and deep. Research before a decision. Structuring a proposal. Working through a pricing question. These involve real back and forth, and they are where the output is worth the most because the alternative was often not doing the work at all.
Most people only ever run the fast loop, which is why the benefit feels marginal.
What enters each loop
Concretely, so this is not abstract.
- 01Fast loop: rewriting, summarizing, formatting, explaining, naming, checking your reasoning
- 02Weekly loop: drafting recurring messages, repurposing content, preparing for meetings, cleaning up notes into something usable
- 03Project loop: research, structuring, comparing options, pressure testing a decision you have already half made
And what does not enter any of them: sending anything without reading it, making a decision, anything a customer will experience as personal, and anything where being wrong is expensive and checking is hard.
The step that is not optional
Everything produced gets read before it goes anywhere.
Not skimmed. Read, with the specific question of whether it is true and whether it sounds like you. A workflow that skips this is not a workflow, it is a way to distribute confident errors faster than you could produce them manually.
This is also the step that improves everything else. What you change on the way out is the signal for what to add to the context document, which means each correction makes the next output need less correction.
You draft a quote follow up, and the output uses a phrase you would never say. You fix it. Then you add one line to the context document: "I never use the word solutions, and I do not open with I hope this finds you well." The next twenty follow ups need less editing. Six months in, most output arrives usable, not because the model improved but because you wrote down what you already knew.
Building it over four weeks
All at once does not work. This does.
Week one. Write the context document. Nothing else. Two pages, in one sitting, imperfect.
Week two. Pick one recurring task you dislike. Run it through the fast loop every time it comes up. Note what you edit.
Week three. Add the weekly batch. Choose one afternoon, collect the week of similar tasks, do them together with the context document pasted in once.
Week four. Update the context document with everything you have been correcting. This is the step people skip and it is the one that makes the previous three compound.
After that, add the project loop when a real decision arrives. Do not construct it in advance.
What breaks it
Four things, all common.
Rebuilding context every time. If you are re-explaining your business at the start of every conversation, you have not built a workflow. Paste the document.
Automating the wrong end. Automating the sending rather than the drafting. The drafting is mechanical and the sending is a judgment.
Treating the output as finished. It is a first draft by a competent stranger who has never met your customer.
Tool switching. Moving between products every few weeks means never learning any of their failure patterns. Pick one, stay a year, learn how it goes wrong.
What this actually gets you
Not ten times output. That claim is made by people selling something.
What it gets a business of one is that the administrative middle stops consuming the week. The follow ups go out. The notes become actions. The research you kept postponing gets done because it takes twenty minutes rather than an afternoon.
The work that requires you specifically still requires you specifically. Nothing here changes that, and anybody telling you otherwise has not run a business where the relationship was the product.
What changes is how much of your attention is available for it.