How to Make $60K/Year Selling Claude AI Second Brain as a Service
Don't want to figure this out alone? I walk members through every step inside the community. Join the Skool → skool.com/raycfu
Every agency, consultancy, and knowledge-heavy business has the same problem. Years of valuable information scattered across Slack threads, Google Drive folders, email chains, call transcripts, project docs, and random notes apps. The people who work there know the information exists somewhere. They just cannot find it when they need it.
They have tried organizing it themselves. They start a wiki. They create a shared drive structure. They buy a project management tool. It works for a few weeks. Then maintenance piles up. Nobody updates the wiki. The shared drive becomes a mess again. They go back to searching Slack and hoping for the best.
The reason every knowledge management system fails is always the same: humans have to do the organizing. That is boring, repetitive work that nobody prioritizes when there are clients to serve and projects to deliver. The AI fixes this permanently because the AI does the organizing. It never gets bored. It never skips a week. It can touch 15 files in one pass without complaining.
You are selling the solution to a problem every knowledge worker has felt but nobody has fixed.
WHAT YOU ARE ACTUALLY SELLING
You are building clients a personal AI knowledge base using Claude Code and Obsidian. The system has three parts.
A raw sources folder where the client dumps everything. Articles, meeting transcripts, client notes, research, strategy docs, competitive intel, call recordings transcribed to text. Everything goes in. Nothing needs to be organized. That is the AI's job.
A wiki folder where Claude writes and maintains the organized version. Summaries, concept pages, entity pages, cross-references, an index, and a changelog. The AI owns this folder entirely. It creates pages, updates them when new sources arrive, flags contradictions, and maintains connections between related topics. The client reads the wiki. The AI writes it.
A schema file that tells the AI how to structure everything. This is the configuration that makes the output specific to the client's business instead of generic. It defines what topics to focus on, how to organize pages, what conventions to follow, and what workflows to run during ingestion, querying, and maintenance.
When a client adds a new source, the AI reads it, extracts the key ideas, writes summary pages, updates the index, and connects it to every related topic across the wiki. One source can touch 10 to 15 pages. The client asks a question against the wiki and gets answers with citations pulled from everything they have ever saved. Good answers get saved back into the wiki so every question makes the next answer better. That is the compounding loop that makes this system valuable over time.
The client never writes the wiki themselves. They never organize anything. They add sources and ask questions. The AI does everything else.
WHO TO SELL THIS TO
Start with agencies and consultants. They are the easiest first clients for three reasons.
First, they have the most scattered data. Years of client work, strategy documents, campaign results, meeting notes, vendor evaluations, and competitive research spread across a dozen tools. They know their knowledge is a mess because they waste time searching for things every single day.
Second, they have the budget. A marketing agency billing $15K to $50K per month per client is not going to blink at $500 a month for a system that saves their team hours every week. The ROI is obvious and immediate.
Third, they already understand the value of organized information. They sell strategy and expertise. A system that makes their expertise searchable and compounding is directly tied to the quality of their work.
After agencies, expand to these verticals:
Law firms. Lawyers drown in case research, precedents, and client files. A searchable knowledge base that cross-references everything and surfaces relevant precedents automatically is worth thousands a month to any firm doing complex litigation.
Real estate teams. Market research, comp data, neighborhood intel, client preferences, transaction history. A real estate team with 5 years of deal data in a searchable wiki can answer client questions in seconds instead of hours.
Financial advisors. Market analysis, client portfolios, regulatory updates, investment research. A knowledge base that flags when new research contradicts an existing position is genuinely valuable for compliance-heavy industries.
Coaches and course creators. Years of content, frameworks, client case studies, and methodology scattered across Google Docs and Notion. A searchable wiki of everything they have ever taught makes content creation and client delivery dramatically faster.
Any business where the value is in what the team knows, not what the team makes, is a potential client.
HOW TO FIND YOUR FIRST CLIENTS
Method 1: Cold outreach via LinkedIn and Apollo
Go to LinkedIn Sales Navigator or Apollo and search for marketing agency owner, SEO agency founder, business consultant, management consultant, or strategy consultant. Filter for companies with 2 to 20 employees. These are the ones drowning in scattered data but too small to have built a real knowledge management system.
Send a simple message:
"I build AI knowledge bases that turn your Slack threads, call transcripts, and Google Docs into a searchable wiki your whole team can use. It updates itself automatically whenever you add new information. Want me to show you a demo using a sample of your own data?"
The key phrase is "using your own data." Generic demos do not convert. When someone sees their own messy notes turned into an organized wiki with connections they never saw, they do not need convincing.
Method 2: Freelance platforms
List on Upwork and Fiverr as an AI automation service. Create a gig specifically around "AI knowledge base setup" or "AI-powered business wiki." Price the initial build at $1,500 to $2,500 on platforms. Once you deliver and the client sees the value, convert them to a monthly retainer off-platform.
Method 3: Communities
Post in Facebook groups and Reddit communities where agency owners hang out. Share a case study or a before-and-after of a knowledge base you built (use your own as the example if you do not have a client yet). Do not pitch directly. Share the result and let people come to you.
Method 4: Local networking
If you are in a city with a business community, attend startup meetups, agency meetups, or coworking space events. The pitch is simple: "I help businesses turn their scattered data into a searchable AI knowledge base." Anyone who has ever searched Slack for 20 minutes trying to find a decision from last quarter will immediately understand the value.
The first client is always the hardest. Build your own knowledge base first as your proof of concept. Use it for a week. Screenshot the wiki, the index, the cross-references. That becomes your demo material. After the first paying client, referrals and case studies do the selling for you.
WHAT TO CHARGE
Setup fee: $1,500 to $3,000 one-time
This covers the initial build. You take the client's existing data (export from Slack, Google Drive, email archives, call transcript folders, project docs), load it into the raw sources folder, write a schema file tailored to their business, and let Claude compile the wiki. The schema file is where the real value is. A generic schema produces generic output. A schema written specifically for a marketing agency that tracks campaign performance, client strategies, and competitive positioning produces output that feels like it was built by someone who understands their business. Because you wrote it that way.
The build takes you 3 to 5 hours for a standard client. As you do more of these, you will develop schema templates for different industries that cut the time down to 1 to 2 hours. You are charging for the result, not the hours.
Monthly retainer: $300 to $500/month
This covers ongoing maintenance. Every month you:
Ingest new sources the client sends you (articles, transcripts, new research, updated strategy docs) Run a health check to catch contradictions, orphan pages, outdated claims, and missing connections Fill knowledge gaps by identifying topics that are referenced but have no dedicated page Update the schema if the client's business focus shifts Deliver a monthly report showing what was added, what was updated, and what the wiki's coverage looks like
The retainer is where the business works financially. The setup fee pays for your initial time. The retainer is recurring revenue with minimal ongoing effort because most of the maintenance is running Claude prompts you have already written.
Vault stacking: $1,000 to $1,500/month per client
This is the upgrade path. Instead of one knowledge base, build three:
Competitive intel vault. Everything the client collects about competitors. Pricing changes, feature launches, hiring patterns, marketing campaigns, press coverage. Every new data point gets integrated into an evolving competitive picture automatically. The client asks "what has competitor X done in the last 90 days" and gets a comprehensive answer with sources.
Client knowledge vault. Everything about every client relationship. Meeting notes, preferences, project history, feedback, contract terms, key contacts. When a team member needs context on a client they have not spoken to in months, they ask the vault instead of searching email for 30 minutes.
Content vault. Every piece of content the team has ever produced. Blog posts, social media, email campaigns, case studies, white papers, presentations. When the team needs to write something new, they ask the vault what they have already covered, what performed best, and what gaps exist.
Three vaults running in parallel takes the retainer from $300 to $500 up to $1,000 to $1,500 per month. The client pays more because they are getting dramatically more value. Three searchable AI-maintained knowledge bases covering their market, their clients, and their content is a competitive advantage nobody else in their industry has.
HOW TO BUILD IT (THE ACTUAL PROCESS)
This is what you do for each client.
Step 1: Collect their data. Ask the client to export everything they can from Slack (use Slack export), Google Drive (download as zip), email (export relevant threads), and any other tools they use. Also ask for call transcripts, meeting notes, and any internal documents they consider valuable. Do not ask them to organize it. Tell them to dump everything into one folder and send it to you.
Step 2: Create the vault. Set up an Obsidian vault (or just a folder structure if the client does not use Obsidian). Create the raw-sources folder and the wiki folder. Drop all their data into raw-sources.
Step 3: Write the schema file. This is the most important step. Create a CLAUDE.md file in the root of the vault. This tells Claude how the wiki should be structured, what topics to focus on, what conventions to follow, and what to do during ingestion, querying, and maintenance. Customize this entirely for the client's business. A schema for a marketing agency looks very different from a schema for a law firm.
Step 4: Run the initial build. Point Claude Code at the vault and tell it to read everything in raw-sources, compile a wiki following the rules in the schema file, create an index, and log everything. Walk away and let it work. When it finishes you have a wiki full of organized pages with cross-references, summaries, and an index.
Step 5: Review and refine. Read through the wiki yourself. Check for errors, weird categorizations, or missing connections. Run a health check prompt. Fix anything that looks off. This quality pass is what separates a professional delivery from a sloppy one.
Step 6: Deliver and train. Walk the client through the wiki. Show them how to add new sources (just drop files in the raw-sources folder). Show them how to ask questions against the wiki. Show them the index and the cross-references. Record a 10-minute Loom walkthrough they can reference later.
Step 7: Set up the retainer workflow. Create a recurring reminder to run the monthly ingestion, health check, and gap analysis. This takes 1 to 2 hours per client per month once you have the prompts dialed in.
SCALING TO $60K AND BEYOND
10 clients at $500/month retainer = $60K/year 10 clients at $1,000/month (with vault stacking) = $120K/year 20 clients at $1,000/month = $240K/year
The system is the same for every client. The folders are the same. The prompts are the same. The workflow is the same. You just swap in different data, write a different schema file, and the AI builds a different wiki. Your marginal cost per client is almost zero once you know the process.
Setup fees on top of retainers add another $15K to $60K per year depending on how many new clients you onboard.
Operating costs are minimal. An Obsidian license is free. Claude Pro is $20 a month. Claude Max at $100 a month if you are running heavy builds. That is your entire tool stack.
The gross margin on this business is 90% or higher. You are selling organized information built by AI. The client pays for the result. The AI does the work. You manage the process.
WHAT MAKES THIS DEFENSIBLE
Once a client has 6 months of knowledge in their wiki with thousands of cross-references and a compounding history of queries and answers, switching costs are extremely high. The wiki is not something they can rebuild with a different provider in an afternoon. It is a living system that gets more valuable every month. That is why retention rates for this kind of service are so high. The longer the client uses it, the more they depend on it.
Build your first knowledge base for yourself this weekend. Use it for a week. Then build one for your first client. By the time you have delivered 3, you will have the process down to a few hours per build and a couple hours per month for maintenance. That is a business.
You just read the full playbook. Most people will close this tab and never implement it. The ones who do usually hit a wall around the technical setup and quit.
Inside the Skool, I walk you through the exact build step-by-step, troubleshoot your setup live in the community, and share the scripts and templates I use to actually land paying clients.
If you want the shortcut instead of the long way around: