Most six-figure experts, coaches, and consultants face this choice: "Should I hire a VA or use AI?"
The answer is both. And neither.
A lot of experts spend $60K-$120K a year on virtual assistants just to find they're still explaining everything, still reviewing everything, and still managing someone else's learning curve.
Others throw ChatGPT at random problems and get boring results because they never built real systems around it.
What works instead: build a $200/month AI system and spend one afternoon training it to work like you.
Six-figure coaches and consultants are using this approach now. Not because it's trendy. Because it actually returns money.
Why $60K Assistants Fail
It's not because the assistant is bad at their job. It's because you can't just hand off something you haven't documented yet.
Here's how this works in most cases:
You hire someone. You explain your process while juggling everything else. They catch 30% of it. They start guessing on the rest. Then you spend 10 hours a week reviewing their work anyway.
You've paid them a salary. But you're still in the loop.
The AI approach is different because the system has to be explicit from day one. You can't delegate to an AI without documenting your process.
And here's the irony: that documentation becomes your first win. You finally know what you're actually doing.
The $200 Stack (What You'll Need)
This is the complete tool list. Total cost: $32.50/month. Setup time: 3 hours.
Claude Pro ($20/month)
What: AI that reasons through complex problems
Why I picked it: Better than ChatGPT at understanding what you actually mean
You could also use: ChatGPT Pro ($20) works fine, but takes more hand-holding
Loom ($12.50/month)
What: Record videos of your screen
Why I picked it: One pre-recorded video replaces 20 minutes of explaining on a call
You could also use: Free Vidyard tier works, but Loom's easier
Typeform (free tier)
What: Pretty intake forms that ask the right questions
Why I picked it: Forms that actually learn as you go, not just generic surveys
You could also use: Google Forms is free but feels basic
Notion (free tier)
What: Your operating manual in one place
Why I picked it: Super easy to organize, find, and share
You could also use: Google Docs works, but Notion is cleaner
Otter.ai ($8.99/month) -- Optional
What: Turns call recordings into readable text automatically
Why I picked it: You get a searchable record of every smart thing you've ever said
You could also use: Free tier catches highlights, paid is worth it
Email service you already have
What: Automated follow-ups and sequences
Cost: Nothing extra—you're already paying for this
Examples: Substack, ConvertKit, ActiveCampaign
That's it. Six tools for less than the cost of one assistant meeting per month. And they work together instead of against you.
The System: How These Actually Work Together
This isn't "use AI tools." This is a workflow.
Phase 1: Intake (Before They Even Call)
Client fills out Typeform with five questions (48 hours before call).
1. What specific outcome would make this investment worth 10x the cost?
2. What have you already tried? Why didn't it work?
3. If we solve this in 90 days, what becomes possible?
4. What happens if this problem isn't solved in 6 months?
5. Walk me through a specific example. Be detailed.
You get 85%+ response rates. This alone eliminates discovery questions on the actual call.
Time: 5 minutes to review responses
Phase 2: AI Analysis (Before the Call)
You copy their Typeform responses into Claude. Use this prompt:
Based on this intake, give me a 150-word brief with:
- The real problem (not the surface symptom)
- The hidden cost they haven't calculated
- The secondary issue this creates
- Their likely objection
- What success would look like in their words
Intake responses:
[paste their responses]
Format for quick reading. I have 90 seconds.
Claude returns a brief in 90 seconds. You read it in 2 minutes. Now you know more about their business than most consultants learn in an hour-long call.
Time: 3 minutes
Phase 3: You Build Your Cheat Sheet (Before the Call)
Spend 5 minutes writing a one-page brief for yourself:
3 questions you want to ask
1 diagnostic framework you'll use
2 solutions you're already thinking about
Their stated problem vs. their real problem (from Claude's analysis)
This is literally your brain on the page before your brain gets tired from talking.
Time: 5 minutes
Phase 4: The Async Proof (Before the Call)
Record a 2-minute Loom:
"Hey [name], I reviewed your intake. Here's what I'm seeing. [Screen share your one-pager]. This tells me [framework]. What this means is [diagnosis]. Here's how we'd approach this: [three-step outline].
Looking forward to our call Thursday where we'll dive into [specific area].
If anything here lands different than how you're thinking, let me know."
Send this email 24 hours before the call.
What happens: When the prospect watches this, they show up already educated and believing you can help. The call shifts from convincing to logistics.
Time: 7 minutes
Phase 5: The Call Itself
The call is now 30 minutes instead of 60, and you close at higher rates because they're already 70% sold. Your only job is to confirm fit, discuss investment, and set expectations.
Time: 30 minutes
Phase 6: Content Extraction (After the Call)
You now have a recorded call. Let Otter.ai transcribe it automatically.
Copy the transcript into Claude with this prompt:
From this call transcript, extract:
- 3 key insights I shared
- 2 frameworks I mentioned
- 1 diagnostic question I asked
- 5 social media posts (100-150 words each, in my voice)
Format each for immediate use on LinkedIn/Twitter.
You get 5 pieces of content. You edit for 10 minutes. You schedule for the next week.
One call = one week of content. Four calls per month = one month of content done.
Time: 20 minutes including editing
The Real Output: What This Stack Replaces
Let's track the time recovery:
Typical week without systems:
4 discovery calls @ 60 min each = 240 minutes
Proposal writing @ 3 hours = 180 minutes
Follow-up emails (manual) = 60 minutes
Content creation = 180 minutes
Admin/scheduling = 120 minutes
Total: 780 minutes (13 hours/week)
With this system:
4 calls @ 25 min each = 100 minutes
Proposal writing = 0 (standardized offers only)
Follow-up (automated) = 0
Content creation = 20 minutes (extraction only)
Admin/scheduling = 15 minutes (automated)
Total: 135 minutes (2.25 hours/week)
You've recovered 10.75 hours weekly. That's worth $1,612/week if you're billing at $150/hour.
Per year, that's $83K in time you got back. Your $200/year tool investment paid for itself in about two days of your recovered time.
The Training Part (This Is Critical)
You have to train Claude on your methodology once. Not repeatedly. Once.
Create a prompt library in Notion. Here's what goes in:
YOUR_DIAGNOSTIC_FRAMEWORK
[Insert your 3-step diagnostic approach]
YOUR_SOLUTION_APPROACH
[Insert your 5-step implementation framework]
YOUR_COMMUNICATION_STYLE
[Paste 5 client emails you're proud of]
YOUR_CONSULTING_NICHE
[Insert your specific expertise area]
YOUR_COMMON_OBJECTIONS
[List the 5 objections you always get]
Every prompt you give Claude references this library:
"Using my diagnostic framework [reference library], analyze this intake and..."
Claude learns your approach, so your follow-ups sound like you and your analysis matches your thinking. That's the difference between generic AI and AI that actually works for you.
What NOT to Automate (Important)
This system doesn't replace you for:
Strategic thinking
Building trust with clients
Problem-solving (the hard thinking part)
Closing deals
What AI actually does: gathers context, organizes information, spots patterns, pulls content, handles follow-up logistics.
You handle: strategy, judgment, relationships, the sale.
A VA costs $60K for this work. You're doing it for $200 because AI doesn't need management, doesn't have bad days, and doesn't need a learning curve.
The Setup Sequence
Week 1: Claude Pro, Notion library creation, basic prompts
Week 2: Typeform intake setup, test with 2 prospects
Week 3: Loom setup, record first pre-call video
Week 4: Otter.ai transcription, content extraction workflow
Week 5: Email automation (onboarding sequence)
Week 6: Full system running
After week 6, you have a system that handles 80% of busywork.
Most experts, coaches, and consultants take 3-4 weeks to see time recovery. By week 8, they're operating at a new capacity level.
One Action This Week
Do this today:
Sign up for Claude Pro ($20)
Create a Notion page called "MY_METHODOLOGY"
Paste your diagnostic framework into it
Paste a successful client email into it
Write one intake question in a Google Form
That's it. You've started the system.
Next week, send that Google Form to your next prospect. See how much you learn from their responses before you even call.
You'll immediately understand why this works better than hiring an assistant.
FAQ
Q: Will this make my work sound generic? A: No. You're training Claude on your actual methodology. It learns how you think, so the output sounds like you because it's literally based on you.
Q: What if Claude gives bad answers? A: It won't, because you've trained it. Bad outputs come from bad instructions. Get the prompt right once, adjust it based on the first result, and you're good.
Q: Won't this take forever to set up? A: Nope. Three hours total. Two hours to build your Notion library with your stuff, one hour to test the Claude prompts. Then it just runs.
Q: Can I use ChatGPT instead of Claude? A: Yes. Claude is better at reasoning, but ChatGPT works. Test both. Go with what you prefer.
Q: What if I already have a VA? Should I fire them? A: No. Use this for the repetitive stuff (intake, follow-ups, scheduling). Free your VA up to do the stuff that actually needs a human — real relationship building, custom work, strategy.
Q: Won't this take forever to set up? A: It takes 3 hours of setup. You'll recover that in your first week of operation.
The $60K question isn't whether to hire an assistant. It's whether you've systematized your process enough that an AI can handle it. Build the system first. Everything else follows.