How Karen runs Oshun Cosmetics
Oshun Cosmetics is a Jamaican medspa and wellness clinic at 9 Devon Road, Kingston, running its full customer journey on Karen AI — IrieVybz AI's sovereign business assistant. The timeline below shows a real 24-hour window: voice calls, WhatsApp, Instagram, bookings, email, post-procedure care plans. Tap any card to see exactly what Karen did and which IrieVybz AI pricing tier includes it.
What is the Oshun × Karen AI case study?
The Oshun case study documents how Karen AI — IrieVybz AI's sovereign business assistant — runs the full customer journey for Oshun Cosmetics Limited, a medspa and wellness clinic at 9 Devon Road, Kingston, Jamaica. Oshun has been a paying AI Studio client ($500/mo plus a 10% revenue commission alignment) since early 2026. Karen handles inbound voice calls through Twilio + ElevenLabs, WhatsApp and Instagram messages, online bookings, post-procedure care plans, weekly performance briefings to the owner on Telegram, and ad creative production via HeyGen. The 24-hour walkthrough below shows the live work.
Specs — Karen AI at Oshun Cosmetics
The deployment in plain numbers.
| Client | Oshun Cosmetics Limited — Kingston, Jamaica medspa & wellness clinic |
|---|---|
| Location | 9 Devon Road, Kingston, Jamaica |
| IrieVybz AI tier | AI Studio ($500/mo) + 10% revenue commission alignment |
| Subscription start | Early 2026 — live, paying client |
| Voice agent | ElevenLabs Conversational AI (agent_1201kj7q18bhfkts8hv2fr3rx3ze) |
| Phone line | Twilio SIP, +1 (876) 676-6297 |
| Channels Karen runs | Voice, WhatsApp, Instagram (DMs + comments), Facebook, email, web chat, Telegram (owner side) |
| Booking surface | Native DB-backed bookings (no Cal.com) — voice/chat → DB row → BullMQ jobs → confirmation |
| Care & retention | 24-hour pre-appointment prep, post-procedure check-in, day-5 retention nudge, weekly owner brief |
| Ad creative | HeyGen Avatar IV (spokesperson "Kesha"), Blotato auto-scheduler, owner-approved via Telegram |
| Chatwoot inbox | Inbox 102279 (omnichannel unified) |
| Setup time | 3–5 days for full content-voice training, Studio onboarding |
| HIPAA posture | BAA-signed configuration, redacted audit logs, per-tenant data isolation, BAA-covered inference path |
| Timeline cards in this case study | 22 scenes across a real 24-hour window |
Why Karen at Oshun, not a generic medspa chatbot
Six specific differentiators between Karen at Oshun and the off-the-shelf medspa booking tools or chatbots the clinic evaluated before going live.
- Voice + chat + email run on one brain. A customer who DMs at 11pm and calls at noon is recognized as the same person. No CRM stitching required.
- HIPAA-ready configuration on a BAA-signed stack. Clinical conversations stay inside the HIPAA boundary with redacted audit logs and per-tenant encrypted Postgres — not a generic chatbot bolted to a clinic.
- Native DB-backed bookings, not Cal.com. Bookings are written directly to the tenant's tables; BullMQ jobs handle reminders. No third-party calendar SaaS in the critical path.
- Owner-approved ad creative pipeline. Karen drafts the spot in Oshun's voice, generates the spokesperson via HeyGen, pings Dr. Rockhead in Telegram for a single-tap approval, then schedules via Blotato. End-to-end in hours, not weeks.
- Care-plan generation tied to procedure type. A smart procedure matcher (regex + Gemini fallback) selects the right care-plan template per booking — Botox, fillers, facials, body treatments — so post-procedure follow-up is procedure-aware, not generic.
- Revenue commission alignment. 10% of attributable AI-driven revenue is shared with IrieVybz AI on the Studio tier. Incentives are aligned to outcomes, not software seats.
Who else this configuration fits
Five audience + scenario pairs where the Oshun setup is a near-drop-in template.
- Solo aesthetic practitioner running their own clinic — operator who is also the practitioner. Karen replaces the part-time receptionist for after-hours and weekend coverage.
- Multi-practitioner medspa juggling Instagram inquiries — clinics where the Instagram comments and DMs are the primary lead source. Karen handles intake, qualification, and booking without missing a 1 a.m. comment.
- Wellness studio with package and series pricing — businesses where bookings repeat (10-session laser package, 3-session facial series). Karen tracks the package state and nudges for the next visit at the right time.
- Clinic with care-plan follow-up obligations — operators who need post-procedure check-ins at specific intervals (24h prep, 2h reminder, day-5 retention). Karen owns the cadence.
- Owner who runs ads but hates production — operator who knows what works on Instagram but doesn't want to film, edit, or schedule. Karen drafts the spot, generates the spokesperson, books approval, and posts.
Customer reviews — Oshun × Karen AI
IrieVybz AI is a new operator-first platform. Public attributed testimonials are being collected — until they're signed and live, the proof is the live deployment itself.
Proof in lieu of public testimonial: Oshun Cosmetics is a paying $500/mo AI Studio client since early 2026 with the 10% revenue commission alignment active. The voice agent (ElevenLabs agent_1201kj7q18bhfkts8hv2fr3rx3ze), Twilio SIP line (+1-876-676-6297), and Chatwoot inbox 102279 are independently verifiable.
Sarah sees the Instagram ad
Sarah is scrolling Instagram late at night. A short UGC spot featuring Kesha appears on her feed.
What Karen did
- Wrote the script in Oshun's brand voice using the loaded voice pack
- Generated the Kesha spokesperson via HeyGen Avatar IV from one reference portrait
- Pinged Dr. Rockhead in Telegram for approval — he tapped the ✅ button
- Posted via Blotato auto-scheduler in Oshun's best-performing time slot
The artifact
Tier
Included in AI Studio ($500)
To replicate for a new client
2 days first time, 1 hour recurring · Needs: Brand voice pack, HeyGen account, Blotato connection, Reference portrait photo
Sarah leaves a comment: "pricing?"
Sarah taps the Kesha ad and lands on Oshun's profile. Without DMing, she leaves a public comment under their latest post: "How much for Botox? 😍"
What Karen did
- Karen's Meta webhook fires within 1-2 seconds of the comment going live
- The comment is captured into Karen's database with full thread context
- Karen pulls Sarah's history: first-time commenter, no prior interaction
The artifact

Tier
Included in AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
10 min (covered by Card 3 setup) · Needs: Meta app approved with comments permission
Karen pings the owner: a lead is asking about pricing
Dr. Rockhead's phone vibrates. A single Telegram message: a labelled lead, the customer's question, and a tap-to-copy reply Karen already drafted.
What Karen did
- Detected a public IG comment within seconds of it being posted
- Classified intent as 💰 lead_pricing using the rule-based classifier
- Drafted a reply in Oshun's brand voice with their booking URL
- Pinged Dr. Rockhead's Telegram with Copy / Skip / Open buttons
The artifact

Tier
Included in AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
45 min · Needs: Brand voice pack, Meta app approval, Comment-reply seed file
Dr. Rockhead taps Copy, pastes the reply to Instagram
One tap on the green Copy button copies Karen's drafted reply. Dr. Rockhead taps Open, jumps straight into Instagram at the right comment, and pastes. Total time: under 30 seconds.
What Karen did
- After the owner taps Pasted, Karen marks the comment 'replied' in her queue
- Sarah is added to the lead pipeline tagged 'instagram_comment_lead'
- If the same person comments again, Karen recognises them across the thread
The artifact

Tier
Included in AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
Included with Card 3
Sarah switches to WhatsApp to ask about availability
Sarah saw the reply, liked the tone, and now wants to actually book. She DMs Oshun's WhatsApp: "What times do you have next week for botox?"
What Karen did
- Karen recognises Sarah from her IG comment (same name, same number when shared)
- Pulls full conversation context — she already knows Sarah is interested in Botox
- No introduction needed — picks up where the IG conversation left off
The artifact

Tier
Included in AI Messenger ($150), AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
20 min · Needs: WhatsApp Business API or Twilio number, Chatwoot inbox
Karen replies in 12 seconds with availability + booking link
Twelve seconds after Sarah hits send, Karen replies in Oshun's brand voice, warm and professional with no filler, sharing two specific time slots and a one-tap booking link.
What Karen did
- Reads available Cal.com slots for the next 7 days
- Picks 2 slots near Sarah's timezone preference (evening, since she DMd at 11pm)
- Writes the reply with brand voice loaded from the Oshun pack
- Sends a Cal.com booking link with treatment pre-selected
The artifact

Tier
Included in AI Messenger ($150), AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
Included with Card 5 · Needs: Cal.com integration
Sarah books her Botox consultation
Sarah taps the link, picks Tuesday at 2 PM, and confirms her booking. The Cal.com page is clean, branded, and takes under a minute.
What Karen did
- Cal.com webhook fires the moment Sarah confirms
- Karen logs the booking against Sarah's contact record across all channels
- Triggers the confirmation email sequence (Card 8) and schedules prep and reminder touches
- Queues the 24-hour prep email (Card 11) and the 2-hour WhatsApp reminder (Card 12)
The artifact

Tier
Included in AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
20 min · Needs: Cal.com account, Cal.com webhook configured, Chatwoot contact linking
Sarah gets the booking confirmation email
Three minutes after Sarah books, the confirmation email lands in her inbox. Not a generic Cal.com receipt — a Karen-written email with prep instructions and a value-first tone.
What Karen did
- Triggered Touch 1 confirmation immediately on Cal.com booking webhook
- Pulled the procedure-specific prep block (e.g. 'avoid blood thinners for 48 hours' for Botox)
- Wrote in Oshun's brand voice using the loaded voice pack
- Scheduled Touch 2 (24hr prep) and Touch 3 (2hr reminder) automatically
The artifact
Tier
Included in AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
30 min · Needs: Cal.com integration, Listmonk SMTP, Brand voice pack, Procedure-specific prep templates
Karen pings Telegram: new booking confirmed
One minute after Sarah books, Dr. Rockhead's Telegram buzzes with the booking summary. No app to open, no calendar to check.
What Karen did
- Detected the Cal.com webhook and extracted client name, treatment, and time slot
- Formatted a one-line booking summary: "Sarah J. · Botox · Tue 2 PM"
- Sent the summary to Dr. Rockhead's Telegram with a View Calendar button
- Updated the owner's daily booking tally used in the morning brief (Card 10)
The artifact

Tier
Included in AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
Included with Card 7 setup · Needs: Telegram bot linked to owner account
Morning brief: leads, bookings, hours saved
Dr. Rockhead opens his email over coffee. Karen's morning brief is already there, one screen tall, with the numbers that matter for the day ahead.
What Karen did
- Pulled the previous day's lead count, booking count, and missed-opportunity alert
- Calculated hours saved vs. a manual front-desk workflow
- Flagged any unanswered WhatsApp threads older than 4 hours
- Sent at 9:00 AM local time via the daily-briefing cron job
The artifact
Tier
Included in AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
15 min · Needs: Listmonk SMTP, Daily cron enabled, Brand voice pack
24-hour prep email with treatment instructions
The afternoon before her appointment, Sarah gets a calm, practical email with exactly what to do tonight and tomorrow morning before her Botox visit.
What Karen did
- Fired the 24-hour prep touch from the scheduled queue set at booking
- Loaded the Botox-specific prep template: no blood thinners, no alcohol, no strenuous exercise
- Personalised with Sarah's name and appointment time
- Included the clinic address and a "need to reschedule?" link to reduce no-shows
The artifact
Tier
Included in AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
Included with Card 8 setup · Needs: Procedure-specific prep templates, Listmonk SMTP
2-hour WhatsApp reminder on the day of the visit
Two hours before her 2 PM appointment, Sarah gets a brief WhatsApp from Oshun. One sentence, the clinic address, and a map link.
What Karen did
- Fired the 2-hour reminder touch from the scheduled queue set at booking
- Chose WhatsApp for the day-of reminder because open rates are higher than email
- Kept the message to two lines: confirmation and address
- Logged the send in Sarah's contact timeline for the post-visit follow-up sequence
The artifact

Tier
Included in AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
Included with Card 8 setup · Needs: WhatsApp Business API or Twilio number
Sarah arrives and receives her treatment
Sarah walks into the Kingston clinic, is greeted by name, and has her Botox treatment with Dr. Rockhead's team. Karen is offline for this part.
What Karen did
- (physical appointment — Karen is offline here)
Tier
Not included in any tier
To replicate for a new client
?
Karen drafts the post-procedure care plan
Four hours after the appointment window closes, Karen pulls the Botox post-care clinical template and drafts a personalised care plan for Sarah.
What Karen did
- Detected the appointment window had closed and triggered the post-visit care sequence
- Loaded the Botox-specific clinical care template with Dr. Rockhead's approved instructions
- Personalised the plan with Sarah's name, treatment date, and clinician sign-off block
- Queued the draft for clinical approval before sending to the client
The artifact

Tier
Included in AI Studio ($500)
To replicate for a new client
1 hour first time · Needs: Procedure-specific clinical templates, Care plan approval workflow enabled, Brand voice pack
Dr. Rockhead reviews and approves the care plan
Dr. Rockhead's Telegram pings with the drafted care plan. He reads it, taps Approve, and Karen sends it to Sarah within seconds.
What Karen did
- Sent the care plan draft to Dr. Rockhead's Telegram with Approve and Edit buttons
- On Approve tap, immediately released the plan to the client email sequence
- On Edit tap, would have opened a text prompt for clinician corrections before sending
- Logged the approval event with timestamp for the clinical audit trail
The artifact

Tier
Included in AI Studio ($500)
To replicate for a new client
Included with Card 14 setup · Needs: Telegram bot linked to clinician account
Sarah receives her approved post-care plan
Four minutes after Dr. Rockhead taps Approve, Sarah gets a clear, reassuring email with her full post-Botox care instructions.
What Karen did
- Released the clinician-approved care plan email the moment approval was confirmed
- Formatted the clinical instructions in plain English at the patient's reading level
- Included a direct line to the clinic for any concerns in the first 48 hours
- Scheduled the Day 5 check-in WhatsApp (Card 17) from this trigger point
The artifact
Tier
Included in AI Studio ($500)
To replicate for a new client
Included with Card 14 setup · Needs: Listmonk SMTP, Clinical template approved by Dr. Rockhead
Day 5 check-in: how's the healing going?
Five days after her treatment, Sarah gets a brief WhatsApp from Oshun. One question, no sales pitch.
What Karen did
- Fired the Day 5 check-in from the post-visit sequence scheduled at Card 16
- Kept the message to a single open question: no links, no upsell
- If Sarah replies with a concern, Karen flags it to Dr. Rockhead immediately
- If Sarah replies positively, Karen queues the review request (Card 18) for Day 14
The artifact

Tier
Included in AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
Included with Card 8 setup · Needs: WhatsApp Business API or Twilio number, Post-visit sequence enabled
Karen asks for a Google review
Two weeks after her visit, Sarah gets a warm WhatsApp. Karen references her treatment by name and links directly to the Google review page.
What Karen did
- Fired the Day 14 review request from the post-visit sequence
- Referenced Sarah's specific treatment to make the message feel personal, not automated
- Included a direct Google review link with no extra steps for the client
- If Sarah leaves a positive review, triggers the referral code send at Day 15 (Card 19)
The artifact

Tier
Included in AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
Included with Card 8 setup · Needs: Google Business Profile review link, Post-visit sequence enabled
Karen sends Sarah a referral code
The morning after Sarah's review goes live, an email arrives with a personal referral code and a note of thanks from the clinic.
What Karen did
- Detected the positive review via the review-monitoring webhook
- Generated a unique referral code tied to Sarah's contact record
- Sent a thank-you email with the code and the referral terms in plain language
- Logged the referral code issue for commission tracking if a referred booking converts
The artifact
Tier
Included in AI Studio ($500)
To replicate for a new client
45 min · Needs: Review monitoring webhook, Referral code generation logic, Listmonk SMTP
Karen sends a personalised nudge when Sarah goes quiet
Six weeks pass with no activity from Sarah. Karen detects the silence and sends one well-timed WhatsApp with a relevant offer.
What Karen did
- Detected that Sarah had no bookings or inbound messages in 42 days
- Pulled her treatment history: last visit was Botox, so suggested a complementary skin treatment
- Wrote a single WhatsApp message in Oshun's brand voice with a limited-time offer
- Sent only once per client per 6-week window to avoid feeling like a broadcast list
The artifact

Tier
Included in AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
30 min · Needs: Re-engagement cron enabled, Upsell service map per treatment type
Weekly brief: hours saved, revenue, recommendations
Saturday evening, Dr. Rockhead gets Karen's weekly briefing email. One page of numbers, arrows, and a short list of what Karen recommends for next week.
What Karen did
- Aggregated the week's leads, bookings, revenue, and hours saved vs. manual front-desk
- Calculated week-over-week movement with directional arrows for each metric
- Identified the top performing channel and the biggest missed-opportunity window
- Included 2-3 concrete recommendations for the following week, ranked by expected impact
The artifact
Tier
Included in AI Autopilot ($250) and AI Studio ($500)
To replicate for a new client
20 min · Needs: Weekly-brief cron enabled, Listmonk SMTP, Analytics data connected
Month-end commission statement and tier performance
On the last day of the month, Dr. Rockhead receives Karen's commission statement with a full breakdown of Karen-attributed revenue and the resulting fee.
What Karen did
- Tallied all bookings attributed to Karen-handled leads for the month
- Calculated the 10% commission on Karen-attributed revenue per the Studio tier agreement
- Produced a line-item statement with booking dates, treatment types, and amounts
- Included a tier performance summary showing which Studio features drove the most value
The artifact
Tier
Included in AI Studio ($500)
To replicate for a new client
30 min · Needs: Commission tracking enabled, Stripe or revenue source connected, Month-end cron configured