There are five stages of AI development.
Most of the conversation happening in aesthetics right now — the webinars, the vendor
pitches, the “10 ways to use ChatGPT in your practice” content — is stuck firmly at Stage 1.
Understanding what comes after it is not a theoretical exercise. It is a strategic one.
Here is what each stage actually means, and where the operational opportunity sits forclinics, medspas, and wellness practices paying attention.
Stage 1: Large Language Models (LLMs)
This is ChatGPT. This is Claude. This is Gemini. You input a prompt, you receive an output.
Most aesthetic practitioners who say they “use AI” are using Stage 1. They’re writing
treatment descriptions, drafting patient emails, generating social captions. That is legitimate
utility — but it is the floor, not the ceiling.
The LLM is a tool. You pick it up, use it, set it down. It does not persist. It does not learn your
practice. It does not act on your behalf. Every output requires a human to initiate the
request and implement the result.
This is the stage most of the industry is treating as a destination. It is an entry point.
Stage 2: Agentic AI
This is where the model stops answering questions and starts completing tasks.You assign a goal — not “draft a follow-up email” but “identify every patient who hasn’t booked in 90 days, segment them by last treatment type, and build a re-engagement sequence for each segment.” The agent uses tools, makes decisions, and executes stepsautonomously to reach that outcome.
In practice terms: AI that can pull from your EHR, cross-reference your booking data, anddraft a targeted communication — without you initiating every step. AI that flags a pattern in cancellations before it becomes a retention problem. AI that monitors your review platforms and surfaces response-worthy feedback in real time.
This is where we are right now. The technology exists. The practices that have moved from
LLM prompting into agentic workflows are already running leaner operations with better
visibility than those still copy-pasting into ChatGPT.
The barrier is not the technology. It is knowing what to build.
Stage 3: Multi-Agent Systems
One agent is powerful. A coordinated system of agents is operational infrastructure.
Imagine a stack where one agent continuously monitors your lead pipeline, another
manages your post-treatment follow-up sequences, a third flags compliance documentation
gaps, and a fourth synthesizes your monthly revenue data into a reporting brief — all
running in parallel, delegating to each other, with no human in the loop for execution.
You are not managing tasks. You are managing outcomes.
The infrastructure for this is being built now. The practices investing in clean, connected data architecture today — unified platforms, structured intake, consistent documentation —are the ones who will be able to deploy multi-agent systems when the tooling matures. The practices running on fragmented systems and manual workflows will be doing the equivalent of trying to add a second floor to a house without a foundation.
Stage 4: Artificial General Intelligence (AGI)
AGI is the stage everyone debates and no one fully agrees on.
The definition: AI that matches or exceeds human cognitive ability across any domain — not because it was trained on specific tasks, but because it can reason, learn, and generalize the way humans do. A scientist, an artist, a strategist — not because it was programmed to be, but because it figured it out.
Whether we have already reached a primitive version of AGI is genuinely contested among the people building these systems. What is not contested: it is the explicit target of every major AI lab operating right now.For healthcare-adjacent industries like aesthetics and wellness, AGI represents the point at which AI can genuinely reason about clinical outcomes, not just pattern-match against them. The implications for treatment planning, patient risk stratification, and evidence-based protocol development would be significant.
That stage is not here yet. But Stage 2 and 3 are.
Stage 5: Superintelligence
This is the stage that generates the most philosophical debate and the least operational
relevance to your practice today.
Superintelligence: AI that surpasses human intelligence in every domain. It does not need guidance, it does not need input, and it works on problems humans have been unable to solve — disease, energy, physics at the edges of comprehension.
Whether this stage is a breakthrough or an existential risk is a question serious thinkers disagree on. It is also a question that has no actionable answer for a clinic owner in 2026.
What does have an actionable answer: where you sit right now between Stage 1 and Stage
3, and what it would take to move forward.
The Operational Reality
The competitive gap being created in aesthetics right now is not between practices that use
AI and practices that don’t. It is between practices that are building agentic infrastructure and practices that are still prompting.
Stage 1 is a productivity improvement. Stages 2 and 3 are structural advantages. The difference is the difference between using a calculator and running a financial model.
The window to build that infrastructure at early-adopter cost and early-mover advantage is not permanent.
This is part of the AI in Aesthetics: The Real Stack series on Aesthetically Audrey —operational intelligence for aesthetic practices, medspas, and the investors evaluating them.
© 2026 Audrey Campbell. All rights reserved.