Mapping The AI Adoption Spectrum From No Tech To Foundational Models

Venn Diagram Viewpoint Of Artificial Intelligence Adoption Market Sizing

When I analyze the AI industry, I think of 2 questions:

Where is innovation happening?
Where is investment flowing?

These questions lead me to consider market sizing – understanding the scale and distribution of AI adoption. This, in turn, raises further questions:

What defines "AI-mature" vs. "AI-aspirational" businesses?
Which types of businesses are more or less common?

If you have any questions or want to discuss AI adoption, get in touch.

A venn diagram with reducing number of businessesa as the AI adoption increases.

The AI Adoption Stack

A useful way to visualize AI adoption is as a layered stack, where each level represents increasing AI involvement and investment. Businesses progress from basic software use to building fully customized AI solutions:

1 - Businesses
2 - Businesses That Use Software
3 - Differentiated Business Using Software
4 - Businesses Using AI Tools Internally
5 - Businesses Embedding AI In Their Products
6 - Running Inference On Custom AI Models
7 - Fine-Tune Transformer models
8 - Train New Models
9 - Foundational Models

The Venn diagram is the core of this article, and the examples below illustrate each stage in the AI Adoption Stack.

1 - Businesses

The broadest category includes businesses, whether they use AI, software, or neither.

Businesses that operate entirely without software are rare today, but some low-tech examples include:

Street vendors – A fruit seller operating from a cart, handling all transactions in cash.
Artisan businesses – A potter making and selling ceramics without a digital presence.
Independent mechanics – Local roadside mechanics who manage billing, scheduling, and diagnostics manually.
Small-scale farmers – Farmers relying on manual labor without farm management software.
Door-to-door salespeople – Salespeople relying purely on in-person interactions without digital lead tracking.
Local barbers or tailors – Businesses operating without digital payments, scheduling apps, or online advertising.

These businesses may still engage with software indirectly, such as through suppliers, customers, or banks, but they do not integrate it into daily operations.

2 - Businesses That Use Software

Most businesses use software in some capacity. Examples include:

Retail – Shopify or Square for POS (Point of Sale) and inventory management.
Restaurants – Toast for digital ordering, OpenTable for reservations, or how McDonald's uses Oracle-based inventory forecasting.
Local services – Jobber for scheduling, Housecall Pro for customer management, and QuickBooks for accounting.
Medical practices – Epic for EHR systems and Zocdoc for patient scheduling.
Real estate – Software for property listings and client management.

3 - Differentiated Business Using Software

These businesses gain a competitive edge using software:

Ride-Sharing – Uber develops proprietary routing and pricing algorithms using real-time traffic data.
Fitness – MyFitnessPal applies predictive analytics to create personalized nutrition and workout plans. Fittr offers an online marketplace with diet and exercise tools.
Real Estate – Zillow provides 3D virtual tours through Matterport integration.

4 - Businesses Using AI Tools Internally

Companies leverage off-the-shelf AI tools to improve internal operations and productivity:

Local Services – Bakeries use ChatGPT to create social media posts and customer emails.
Real Estate – Agencies generate visually appealing property listings with Midjourney.
Recruitment – Recruiters rely on HireVue’s AI models for resume screening and interview analysis.
Functions within a business – Teams use Notion AI to summarize meetings and draft project updates. Marketing teams generate content with Copy AI. Legal teams review contracts using Lexion.
Software Companies – Developers rely on GitHub Copilot for code completion and productivity boosts.

88.99% of small and medium businesses (SMBs) report actively using AI in their operations.

5 - Businesses Embedding AI In Their Products

These companies integrate AI into their core offerings using third-party APIs like OpenAI and Anthropic without training their own models:

Perplexity.ai – Provides AI-powered search and conversational answers, competing with traditional search engines through LLMs.
Coso.ai – An AI-driven social media management platform that generates and manages brand-specific content based on social trends, improving engagement and saving time.
Superhuman – Uses AI for email triage, smart replies, and prioritization to optimize workflows.
Canva – Leverages AI for background removal, image enhancement, and Magic Write (text generation) via OpenAI APIs.
E-commerce – Etsy uses Google Cloud AI for personalized product recommendations.

6 - Running Inference On Custom AI Models

These businesses train and deploy their own AI models to run inference at scale for specialized applications:

Fraud Detection – Stripe Radar uses proprietary machine learning models to detect suspicious transactions in real-time.
Autonomous Vehicles – Waymo processes sensor data with in-house deep learning models for object detection, lane following, and decision-making.
Medical Imaging – Aidoc runs AI-powered radiology models to detect anomalies in CT scans and X-rays.
Retail Personalization – Amazon deploys proprietary recommendation models based on user behavior and product embeddings with custom LightFM variants.
Supply Chain Optimization – Walmart Uses forecasting models to manage inventory and reduce waste.

7 - Fine-Tune Transformer models

Businesses fine-tune pre-trained transformer models on proprietary datasets to enhance performance for domain-specific tasks:

BloombergGPT – Fine-tunes financial language models on SEC filings, earnings reports, and market data for specialized financial analysis.
Epic Systems – Adapts large medical language models for electronic health record (EHR) analysis and clinical documentation.
Duolingo – Customizes language-learning AI by fine-tuning transformer models on learner interactions to optimize grammar correction and personalized feedback.
Salesforce Einstein – Integrates transformer-based models into its CRM platform to generate personalized sales insights and automated customer support responses.

8 - Train New Models

Organizations developing AI models with unique architectures or training methods (without using foundation models) to enhance their product’s value to customers:

LG AI Research (EXAONE) - Used Amazon SageMaker to develop EXAONE, a large-scale AI system powering the AI artist Tilda.
E-commerce startup - Built a proprietary model to help merchants edit product photos and videos.
3D modeling startup - Developing a digital twin of the real world with generative elements.
AstraZeneca - Uses AI to accelerate drug discovery research.
DeepMind (AlphaFold) - Develops state-of-the-art models for protein folding prediction, reinforcement learning, and multimodal AI reasoning.
Adept AI - Trains action-based AI models to automate software workflows beyond text-based interactions.

9 - Foundational Models

Organizations developing general-purpose AI models designed for broad use and fine-tuning:

Anthropic (Claude AI) - Focuses on alignment-first LLMs, leveraging Constitutional AI for safety and interpretability.
Mistral AI - Specializes in lightweight and mixture-of-experts architectures, optimizing for efficiency and open-weight availability.
OpenAI (GPT-4, Sora) - Builds general-purpose LLMs and multimodal models for enterprise AI applications.
Google DeepMind (Gemini) - Develops multimodal foundational models capable of understanding text, images, and code, enhanced by reinforcement learning.
Meta (Llama 3) - Focuses on open-weight foundational models optimized for efficiency and decentralized deployment.
Hugging Face (BLOOM) - Hosts and maintains open-weight foundational models but does not regularly train new ones.

Ending Thoughts

The above examples can help you understand the venn diagram in this article.

AI adoption isn’t just about technology—it’s about competitive advantage. Companies that recognize where they stand in this stack can make better investment decisions.

Where does your business fit today, and where will it be tomorrow?

If you have any questions or want to discuss AI adoption, get in touch.