AI Economics

The business case for GenAI

Martin Casado and Sarah Wang from a16z published a piece recently entitled The Economic Case for Generative AI and Foundation Models.. It’s a thoughtful treatise that we believe will endure the test of time, addressing the economic challenges with AI historically and currently. The ‘cliff notes’ version of the article is that outside of the few big players in the industry, it is difficult for companies to create high-margin, high-growth business centered on traditional ai due to a ‘mediocrity spiral’. In other words, a company tends to get locked into a cost structure that is difficult to unwind as the business scales, and therefore the high returns never materialize.

But something different is happening with Generative AI, a new class of intelligent processes built on an underlying technology that can do more than generate text. That’s what has riveted our focus and has been at the heart of the protypes we’ve built for clients. While the wizardry of language models is interesting, the power of these same models to process logic is captivating. We know this angle has the attention of a lot of companies as well, given the extraordinary rate of adoption of GPT. However, we've observed that companies deploying the technology internally are still struggling to capture the productivity gains.

Let's delve further into the business case for generative ai. If a company's focus is content creation, the economic principles at play are readily understood. Experts like Casado and Wang have noted significant shifts in marginal costs across various sectors. Microchips have essentially reduced the marginal cost of computing to zero, while the internet has similarly impacted the marginal cost of distribution. GenAI takes it one step further by effectively reducing the marginal cost of content creation to zero. The economics of the enterprise is changed for any business which specializes in advertising copy, marketing content or graphic design.

However, while the content-creation side of GenAI is impressive, it is not the ‘needle moving’ activity needed by most businesses. We’ve written on this topic from the outset of GenAI here and here.

So how do we structure the business case for GenAI? We believe that logic models represent a ‘wave 3’ opportunity for most companies, where core business processes can be reengineered radically to leverage the ‘invisible engines’ of GenAI for various tasks. The key is analyzing a core process and identifying the collection of activities and tasks that comprise the process. For any sequential set of tasks, or elliptical tasks in the event of a knowledge-based process, the workflow may be restructured to delegate work to and accelerate outcomes through the foundation model. In a software development process, for example, code reviews are no longer handled by senior developers but conducted by the foundation model with greater precision and speed. Exception handling and returns are restructured in a supply chain process to draw on the powerful logic engine of GenAI, sorting through myriad transactions and consuming apis to complete processes at scale.

In our view, the business case is more than just the marginal cost of another unit of production, but a wholesale opportunity to change the competitive posture of the firm in costs, quality, speed and risk.

Give us a call so we can discuss setting up a prototype for one of your core processes.