The most common reason experts don't package their knowledge is not lack of interest — it's paralysis in the face of a blank page. The task feels enormous. Everything you know is connected to everything else; where do you even start? The answer is: you don't start with everything. You start with the questions.
Step 1: List the 15 questions you get asked most often
If you've been working in a domain for more than a few years, you have a set of questions that come back reliably — from clients, from people on forums, from colleagues, from friends who introduce you as "the person who knows about X." Write those questions down. Don't filter them. Don't decide which ones are interesting enough. Write down the real ones, including the ones that feel too basic.
These questions are valuable precisely because they're common. The people asking them need real answers, not generic ones. Your answers — grounded in actual experience — are more useful than anything a search engine will surface. That list of 15 questions is the core of your vault.
Step 2: Write your real answers — don't polish them
For each question, write what you would actually say if someone asked you directly. Not the polished version. Not the "what I'd put in an article" version. The version with the caveats, the edge cases, the "it depends on whether you're in X situation or Y situation," the "here's what most people get wrong about this."
Imperfect, honest answers are more useful than clean, generic ones. The AI grounding the vault will work with whatever you give it — and specific, opinionated content produces far better results than hedged, surface-level content. Aim for 150–300 words per answer. Don't try to be comprehensive. Try to be correct and useful.
Step 3: Add your frameworks and reference materials
What else do you have? Notes from client work, checklist documents, frameworks you've developed, PDFs you reference constantly, case studies from your actual experience. All of it can go into the vault. You don't need to reformat it — paste it in, upload it, add it as-is. More context makes the AI more accurate. The goal is to give the system enough grounding that it can answer questions from your perspective rather than from the generic training data behind it.
Step 4: Write a system prompt that defines your expertise
The system prompt is the instruction set that tells the AI how to behave. Keep it direct: who you are, what the vault covers, what it doesn't cover, and how you want it to handle questions outside its scope.
A working example: "You are an AI assistant trained on the knowledge of [Name], a licensed electrician with 18 years of residential and commercial experience in the Pacific Northwest. Answer questions using only the documents in this vault. For questions about code compliance, reference the specific NEC sections in the uploaded documents. If a question requires a site visit or involves safety risk, say so clearly and recommend consulting a licensed electrician on-site."
Specific beats generic. The more clearly you define the scope, the better the AI stays within it.
Step 5: Set a price and publish
Price based on what an hour of your time is worth to a client. If you bill $200/hour, a vault priced at $49 for 30 days of unlimited questions is a clear value proposition — the buyer gets the equivalent of a multi-hour engagement at a fraction of the cost, and you deliver it without spending the time.
Publish it and share the link. The vault will improve as you add more content over time, but it delivers value from day one. You can start building your vault right now — the setup takes less than an hour, and the first buyer can arrive the same day. If you want to see what other practitioners have built for reference, see example vaults in your category first. The full mechanics are covered in how vault creation works.