Every point from Corey Gannon's "simplest AI business of 2026," extracted whole and mapped onto what Cakesmash already is. The doctor model: diagnose pain, prescribe off-the-shelf tools, open the door to real money. Teal bands are the Cakesmash translation.
Source: Corey Gannon × Greg — YouTube · full 11,925-word transcript pulled & read · every dollar figure below is the presenter's stated number, not a Cakesmash projection.
Where this lane stands
Core finding
Cakesmash already runs this machine, pointed at medical + video. The Vitals Audit is his assessment; the retainer is his implementation upsell. This episode adds a lighter foot-in-the-door, a new recurring-revenue rung (AI Concierge), and 7 zero-capital acquisition plays.
Direction
LeaningKyle Cassie = the horizontal AI-consulting brand (home for this model + BoardSmart + Momo + LA SMBs), under the Cakesmash Media LLC umbrella. Cakesmash stays the medical authority + video brand, unmuddied.
First rep
Run the AI Concierge as the wedge, with rep #1 sourced from outside the current client base (referral partners + network mini-audits). Trust before upsell.
Locked Jul 17
One Smile stays on their current video/reels track. Do NOT pitch them the Concierge yet — keep building trust with our first paying client, revisit a possible pitch ~Sept 2026.
You meet a business owner, spend ~45 minutes finding their biggest time drains, and prescribe AI tools that fix them. Like a doctor writing a prescription. No audience, no capital, no coding. You aren't building or implementing. You prescribe tools that already exist.
Every US business needs AI in some capacity. Roughly 5% use any AI tool beyond prompting ChatGPT. That gap is the whole opportunity.
The market is so wide that giving the entire playbook away leaves opportunity left over. Blue ocean.
Owners are stuck two ways: too buried (40–60 hr weeks) or head-in-sand afraid competitors are pulling ahead. Both are your opening.
Cakesmash translation
You already believe this, it's why the Vitals Audit exists. His version is horizontal and lighter (any SMB, $999); yours is vertical and heavier (medical, $497–997). His model is a faster cash engine that can feed the same retainer. The 5%-adoption stat is even more true in dental/medical, where owners are clinicians first and least likely to have touched AI.
1 The Offer — the "$999 AI Tools Assessment"
ICP: owners of businesses with 2–20 employees, $500K–$5M/yr revenue. The sweet spot.
"We'll sit with you 45 minutes and do a structured interview. We pull out your bottlenecks and pain points, then prescribe 3–7 off-the-shelf AI tools you can implement immediately to reclaim 5–10 hours per week. If we can't find at least 5 hours, you get 100% of your money back. Worst case you lose 45 minutes. Best case you get 5+ hours back every week, forever."
Price: $999. The foot in the door, not the product. Lifetime value lands $3K–$10K+.
Real numbers: ~15 assessments sold in 2026; 50% ask for implementation after.
The money-back guarantee makes it irresistible: the only downside a client can name is 45 minutes.
"A good audit gets you paid to uncover opportunities for the client to pay you more." Put ~$200 of each $999 into ads and it self-liquidates.
Cakesmash translation
Your Vitals Audit is this offer, already priced and medical-tuned. Upgrades to steal: the hard time-based guarantee ("5+ hrs/week back or full refund"), the 3–7 tool prescription format, and the reclaimed-hours headline number. A dental/ortho practice fits his ICP exactly (owner, small staff, high hourly value).
2 The Four-Phase Fulfillment
Phase 1 — Discovery call
Zoom / Meet with an AI notetaker recording (Fathom, Otter, Fireflies).
You only ask questions and pull out problems. No pitching. No prescribing on this call (bite your tongue even when you know the tool).
Question bank: Walk me through yesterday · What tasks do you dread? · Where does work pile up? · What have you tried to automate and failed? · Magic wand: delete any one process, which is it? (answer is very often email).
Phase 2 — Analyze the transcript with Claude
Feed the transcript to Claude: "Here's a call transcript with a business owner. Research off-the-shelf SaaS/AI tools that fix these pain points." Claude catches patterns you missed.
Corey productized this as a Claude Skill improved via evals (feed past transcripts + finished reports back). First 2–3 land ~60–70%; by the 4th–6th it's often copy-paste.
QA is mandatory. Right-size every tool: a 4-person landscaper gets a small-business CRM, not Salesforce.
Research directories: futurepedia.io (acquired by HubSpot), theresanaiforthat.com — filterable by industry. Much of what you prescribe is plain SaaS they didn't know existed. Still a win.
Phase 3 — Generate the report
The client-facing deliverable, built in Claude Design (free template at audittemplate.ai). Governing law: a confused mind doesn't buy, doesn't implement, doesn't upsell. Corey iterated it 12 times. Full anatomy in §3.
Phase 4 — Review call
Email the report before the call. ~30-min Zoom, screen-share, walk each recommendation.
The three closing questions: (1) Which is most urgent? (2) Do you want to do this yourself or would you like my help? (3) What's your timeline — killing you, or could you live 60 days with them?
50–60% choose implementation. That opens the upsell menu.
Cakesmash translation
Phase 2 is a Lucius job you already do daily — I run the transcript, right-size + QA tools, build the report. Operating spec saved as INTERNAL-lucius-assessment-operating-prompt.md. Phases 1 + 4 are Kyle on camera, and a director on the call reads as premium. Standardize on one notetaker.
3 The Report Anatomy
Title — client name, date, business type, primary focus.
Executive summary — 1–2 main pain points · main outcome · hours reclaimed/week (5–10, avg 7) · primary focus = which ROI lever most tools pull.
Effort/Impact matrix (below).
Quick-wins summary — one line each: pain → tool
Recommended solutions — deep dive per tool: name · pain solved · cost · setup time · time saved. ~50% of the review call lives here.
4-day quick-start plan — day 1–4, each action <10 min. Beats the freeze.
What comes after quick wins — the "major projects" needing a build. Seeds the upsell.
Financial impact — the money slide (below).
Next steps — do the 4-day plan + book the review call.
▲ Quick Wins
High impact · Low effort
Off-the-shelf tools. Sign up and go. The report body.
◆ Major Projects
High impact · High effort
Needs a build — agent, knowledge base, skill. The upsell slide.
Low impact · Low effort
Ignore.
Low impact · High effort
Ignore.
Financial impact formula: Monthly net ROI = (hrs saved/wk × ~4.3 wk × client hourly value) − monthly tool cost
Avg tools ≈ $60/mo. Avg time saved ≈ 7 hrs/wk. Owner's time is worth hundreds/hr. Net ROI is always four figures, sometimes five. Anchor it to what they paid: "You paid $999 — here's $X,XXX/month back."
The one meta-takeaway: make the output stupidly simple. Anyone glances at the exec summary and instantly knows their pain, outcome, and hours back. Delete, simplify, rearrange until the ROI is impossible to miss.
Cakesmash translation
Your web-design baseline already beats a generic template on polish. Reuse the Vitals Audit HTML shell, add the effort/impact matrix, 4-day plan, and financial-impact slide as the ROI teeth. Gate #22/#30 stay live: every stat links to source; tool-count and hours must be defensible from the transcript, never invented.
4 The Upsell Menu — Six Rungs
"$999 is just the door." After the assessment these are warm clients, never cold. There's always an upsell to the upsell.
Rung
What it is
Price
Process redesign
Fix a broken process before automating it (16 steps → 7). No AI. Ex: e-comm seller spending 10 hrs/wk on Amazon ad-spend monitoring — fix the process first or automation falls apart.
$3K–3.5K
Automation build
Zapier / Make / n8n. Cut-and-dry input→output, 1–3 steps. Ex: wedding-venue ops manager duplicating Asana projects per client → one Zap.
~$1.5K
Knowledge system
Custom GPT on their material. Ex: business broker got 400–500 emails/listing, same 5 questions — custom GPT on the listing packet, 500 emails → 0. Both sides called it the best experience they'd had.
project
Custom workflows
Claude skills for a proprietary process; bake in a recurring maintenance retainer (processes change every few months).
$3K + retainer
Full implementation
Any combination of the above.
~$8K
AI Concierge
Done-with-you monthly retainer. The recurring-revenue crown jewel (§5).
$1.2K–2K/mo
Retainer-ize any rung: "$3K/mo for up to 2 knowledge systems," "$2K/mo for up to 3 workflows." Sell monthly by proving one build first, then offering more.
Cakesmash translation
All in your wheelhouse; I can fulfill the build work today. Note this menu is horizontal-ops, not video — under Cakesmash-medical the natural upsells stay your reels + AEO + retainer, with these as the operational add-ons that make you stickier than a pure video shop. The AI Concierge is the cleanest new rung regardless of the fork.
5 The AI Concierge Retainer — THE CROWN JEWEL
Done-with-you consulting. Corey makes $1,000+/hr. Told the first 6 people, closed 5, hit $8K MRR in 10 days, raised the price on every yes, they kept saying yes.
The offer
Two 45-min calls/month, client screen-shares, you help them use Claude and build Claude skills. At $1,500/mo that's a $1,000/hr effective rate.
Unlimited Voxer, 12-business-hour response. High perceived value, near-zero real load (of 5 clients, 3 messaged once each in 2.5 months).
Call 1 = foundation only, no building. Guided onboarding plugin: connect tools · context files · global instructions · show a scheduled task · show a skill.
Every call after = AOA:Audit (show me how you do it) → Optimize (cut 13 steps to 7) → Automate (turn it into a Claude skill). Rinse, repeat.
What makes it feel premium
Pre-engagement Google Form (a mini-audit): "90 days from now, what makes this a win?" Then deliver it — 4 of 5 clients hit their 90-day win in ~60 days.
Notion one-pager hub per client: running inventory of everything accomplished, emailed after every call. It justifies the next invoice.
At renewal: "You said success looked like X — we exceeded it." Then upsell again.
Cakesmash translation
Highest-leverage single addition in the whole episode, and brand-neutral. A Concierge at $1,200–2,000/mo is a far lighter sell than a $3–6K video retainer and gives MRR you can stand up this month. I build the onboarding plugin, the context/skills setup, and the Notion hub. First rep comes from outside the current client base — One Smile stays on their video/reels track for now.
6 Pricing Psychology
Credit-the-audit close: "I'll credit the $999 toward the implementation, so it's $4K not $5K." The move: mark the upsell UP by ~$1K first, then credit the audit — you net the same or $1 more, they feel $1K off. The audit becomes effectively free.
Tiered scarcity pricing: "4 spots at $800/mo, then $1,200, then 4 at $1,600." Services need manufactured urgency or nobody moves.
Legitimate scarcity cap: "I'm capping at 6 clients. I have 2 other calls this week — if you pass and one says yes, I can't take you until someone churns." Only works if actually true. Never fake it.
Go up-market over time: drop/wait-out lower clients, build a more premium tier. There's always a more luxurious experience some ICP wants. One step at a time.
7 Seven Ways to Get Clients — no capital, no audience
These win because 99% won't do them, and the ones who try quit in 3 days. Hang in 7 days and you're in the top 0.001%.
1
Local "AI for Business" meetup
Slow burn
Barter a co-working space's conference room free ("I'll bring 30 local owners"). You host = presumed expert. Network 10–15 min, give a 20-min "how to use Claude" talk, collect contacts, follow up within 24 hrs. Clients come by event #3–6 (SEO-like). Make it a recurring monthly event brand where you're the leader.
2
Door knocking
Immediate
Best way to learn sales, full stop. One listener hit 30 businesses → 5 meetings → 2 clients. Need a client today? Knock a door. Nobody does it — that's the moat.
3
LinkedIn DMs (probe, never pitch)
Fast
Not spam slop. Target local owners. No pitch in message one. "Hey Joe, I'm Corey, probably live down the street from you. Curious how you're using AI. Happy to give free feedback." The "live down the street" hook is gold. Use voice notes or Loom.
4
Free mini-audits for your network
Fast
Everyone knows owners (gym, church, other parents). "I'm getting into AI services. 15 min, free, one way AI could help your business — open to it?" The mini-audit is the lead magnet; the full assessment is the upsell. Irresistible framing: "Worst case, you learn one or two tools you'd never heard of. Best case, you get a partner who can build this AI stuff off your to-do list."
5
Agency / professional referral partners
Compounding
Your insurance agent, accountant, marketing-agency owner all have ICP clients asking them about AI (which they can't answer). "Send me anyone with AI questions, I'll pay a referral fee." Key = follow up every 2–3 weeks. Level up with a co-branded workshop ("AI to optimize your finances, brought to you by [you] + [the firm]").
6
AI office hours at a co-working space
Fast
"I'll come in one hour a week as a free AI resource for your tenants." Win-win-win, and you plug straight into a room of your ICP. Real ex: Dennis pitched 10 spaces, got 2 yeses, door-knocked shops for $175 in gift cards as a draw, got 11 people to session #1, 2 follow-up calls booked.
7
Post your wins / build in public
Slow burn
Document as you go — short videos, X posts. A "win" is tiny: "someone texted me a ChatGPT question, here's the great answer." Early on manufacture it: "one valuable prompt a day for 90 days" as carousels; real wins layer on as clients arrive.
The combo play: stack method 5 + 1 — partner with a super-connected local pro (Corey used a plugged-in realtor) to co-host the meetup; got a client from the first one. Method 6 + 2 stacks the same way.
Cakesmash translation
These are tuned for a local generalist — which is exactly the Kyle Cassie / LA-horizontal lane. For a medical-vertical angle the analogs are #5 (dental CPAs, practice-management consultants, supply reps), #4 (mini-audit to One Smile's network), #3 (LinkedIn to practice owners), #7 (your reel wins). The meetup / office-hours / door-knocking plays only fit if you run the LA-horizontal lane — which is why the brand fork above decides which of the seven go live.
8 Niche Down (then niche down again)
Thousands are starting this, but the pie is effectively infinite — winners focus. Be the go-to for one geography ("the AI guy in Charlotte") or one industry ("AI for financial services"). Higher price, far less pricing pressure when expertise backs it.
Cakesmash translation
You have both levers, rare: an industry (medical/dental, locked + a live client) and a geography (LA), plus an edge nobody else has — a working film director. Corey's own advice: pick one and go deep. The Kyle Cassie brand can lead with LA geography now, then earn the dental-industry niche once One Smile teaches you the stack.
9 What I Execute On Your Word
Fork-agnostic, can start the moment you confirm the One Smile question above:
Stand up the AI Concierge offer ($1.2–2K/mo, two calls + Voxer + AOA). MRR this month. I build the onboarding plugin, context/skills setup, and Notion accomplishment hub.
Productize the assessment: turn the operating prompt into a live Claude skill (transcript → right-sized prescription → report), reusing the Vitals Audit HTML shell + effort/impact matrix + financial-impact slide.
Add the time-guarantee + reclaimed-hours headline to the audit offer.
Pick ONE acquisition method to run first (my rec: #5 referral partners + #4 network mini-audits — fastest, warmest, no room to book).