Using ChatGPT does not make an AI company: a guide for investors

The recent advancements in AI have generated substantial interest in the investment landscape, prompting numerous investors to explore opportunities for leveraging the technology’s potential for industry transformation and growth. AI has become an absolute necessity for establishing and sustaining a competitive edge, and we are already witnessing tangible benefits for a few early adopting companies, making it even more urgent for others to accelerate their digital transformations. 

Similar to the late '90s dot-com bubble (though seemingly not as hyped as then), the AI domain has become a hotbed of excitement and innovation. However, as history has shown, not all that glitters is gold. The popularity of AI has also given rise to a troubling trend: companies falsely claiming to have AI integrated into their Intellectual Property (IP), when in reality they are merely end-users of AI technology.

For investors, discerning between companies genuinely leveraging AI and those simply riding the wave of AI hype is crucial in making well-informed investment decisions. Investors should remain vigilant and be able to distinguish between the companies merely echoing buzzwords for marketing purposes and the real deal. Just as with the dot-com bubble which eventually burst, separating the visionary companies from the speculative ones will be the key to long-term success in the AI investment landscape.

To give a clearer perspective, we’ll categorise companies into four distinct classes based on their AI adoption levels and offer guidance on what investors should consider when assessing a venture’s AI capabilities:

While this category may seem surprising for a startup company active in the tech sector today, there are still businesses that rely solely on traditional methods and have not yet explored AI's potential. These companies often operate in industries where AI adoption is not a pressing concern, such as certain manufacturing sectors or small local businesses. They rely on traditional software and processes and may have missed out on the opportunities that AI can offer, such as process optimisation, automation, and improved customer experiences. There are also businesses where there is recognition of AI's potential value, but they have not taken the first steps yet. They are in the phase of evaluating opportunities and planning to take the necessary steps in the near future.

As an investor, it's crucial to determine whether a company's lack of AI integration is a deliberate strategic choice or a reflection of a lack of technological adaptation.

These companies have incorporated basic third-party AI services into their products with minimal or no customisation. This integration typically provides nice-to-have, side features that enhance the overall user experience. They rely on readily available APIs, which allow them to quickly implement AI capabilities such as standard language translation, sentiment analysis, or basic chatbots.

These added capabilities do not contribute any value to the company’s IP as these companies neither develop their own proprietary AI models nor qualify as advanced users of third-party AI services. They solely rely on off-the-shelf AI tools and services.This situation can be a pitfall for investors, as these companies are capitalising on the AI hype without delivering distinctive value to their customers. A knowledgeable user could potentially access these AI services directly without even utilising the company's product.

However, these companies may possess the potential to enhance their integration with AI services in later stages (see the category below: AI enhancer). For investors, it's essential to evaluate the depth and effectiveness of these companies’ roadmap in the AI domain. Consider factors such as the relevance of planned AI features to the core product offering and the potential for these features to drive user engagement and revenue growth in the future.

Moving up the AI ladder, we encounter companies that take AI integration a step further. These companies take a more proactive approach by providing deeper AI services that enhance their product offerings. Instead of relying solely on third-party APIs, AI enhancers involve custom context, drawn from their customers’ data as input and they leverage AI (e.g. via embedding custom data) to create tailored, context-aware experiences for their users.

A Content Management System (CMS) that indexes a company's documents and uses a third-party AI service to provide contextually relevant responses, or an e-learning platform that offers personalized content recommendations based on a student's progress and preferences are examples of an AI enhancer.

These companies do not necessarily develop AI models from scratch, but they provide value-added services by effectively utilising AI technologies to benefit their customers. This enriches the user experience by delivering information that is highly relevant to the user's query.

Investors should assess the scalability and uniqueness of these services, as they can significantly differentiate a company in a competitive market. Furthermore, Investors should evaluate the extent to which AI enhancers leverage data to provide context-aware, tailored solutions. Companies that can efficiently access, manage, and utilise data to enhance AI offerings can deliver highly relevant and context-aware AI-driven solutions which may lead to having a competitive edge.

At the forefront of AI development are the AI pioneers. These companies showcase a deep commitment to AI innovation and often already have a competitive edge in their industry. They either develop and train their own AI models from scratch or heavily customise well-known existing AI models (via techniques like fine-tuning) with proprietary datasets to meet their specific needs. 

AI pioneers often lead the way in innovation, using AI to predict market trends, enhance core product features, or solve complex industry-specific problems. A financial institution which develops its AI model to detect fraudulent transactions or a language translation service that fine-tunes AI models (such as Llama2) for industry-specific jargon and terminology are examples of AI pioneers.

These companies are considered to have the most potential to become successful and turn into industry leaders. Core product features stand on the shoulders of modern AI services developed by in-house expert teams who continuously leverage unique quality data to train and enhance their AI models to provide better solutions for their customers. Investors should carefully assess the company’s model development, data practices, required in-house expertise, and roadmap to ensure that the company has the potential to become the next big star.

Assessing companies’ claims in the AI sector demands a nuanced approach. Several aspects warrant evaluation when considering an investment in a company with AI claims. Below, we outline some crucial considerations:

• In-house AI expertise: Assessing the depth of a company's AI expertise can reveal much about their commitment to pioneering AI. Do they maintain or plan to have a dedicated team of data scientists and AI/Machine Learning (ML) engineers? The presence of a talent pool focused on AI is a strong indicator of the company's dedication to advancing in this field.
• Data strategy: The size and quality of data lie at the core of a successful AI product. The uniqueness and magnitude of a company's datasets are crucial factors for the development of powerful AI features. Without access to quality, extensive datasets, and well-defined data processes, it's nearly impossible to develop robust AI features. Large, proprietary datasets can be a significant asset in AI development, providing a competitive advantage for the company, particularly for AI pioneers. Understanding the company's data assets and its approaches to data acquisition, management, processing, and protection is critical when assessing its AI potential.
• Data privacy/legality: Given the possibility of sharing customers’ data with third-party AI services or dealing with Personal Identifiable Information (PII), understanding the extent of data privacy measures is crucial for investors. Evaluating data handling practices to ensure compliance with privacy regulations is essential as data security is paramount due to the use of customer-specific data.
• Development process and Machine Learning Operations (MLOps): This is relevant to companies which develop their own models (i.e. AI pioneers). A robust AI product should be backed by best practices in model development, training, testing, and provisioning. Assess the company’s model deployment and monitoring mechanisms, ensuring seamless integration and ongoing performance as these factors impact the reliability of AI models. A well-structured and efficient model development process, integrated with MLOps, indicates a commitment to delivering scalable and dependable AI solutions, essential for long-term success in the AI landscape.
• Cost efficiency: Providing AI services demands a significant amount of processing power compared to traditional software. Evaluate the scalability of their AI-driven features and their ability to serve a growing user base. What are the current hosting and operational costs for AI models? What would the costs be if the user base were to increase tenfold? 
• ‍Vendor reliability: If the company relies on third-party service providers for its AI offering, assessing the reliability of these third-party AI vendors is crucial, as the company's performance and product availability directly depends on them.

Different shades of AI integration across industries and sectors have blurred the lines between traditional business practices and the cutting-edge innovations that AI brings. Whether a company is just beginning to dip its toes into AI or is pioneering new AI-driven solutions, understanding their position within the AI landscape is crucial to gauge growth potential and competitive advantage.

Conducting thorough due diligence on AI strategies, industry contexts, and data practices is essential to assess the effectiveness of AI integration. This, in turn, enables investors to make informed decisions in the rapidly evolving AI landscape.