Over the past few weeks, I've been working with several "portfolio" companies to help them understand the impact of large language models (LLMs) like GPT on their product and business strategy. The goal was to figure out how to respond to the challenges and opportunities of the new landscape. Here are some of the key takeaways we've gathered:
LLMs are a big deal. A really big deal. If you think LLMs won't affect your business soon, think again.
Beware of the hype – it shall pass.
Public foundational models (like OpenAI's or Google's) on their own won't give you a sustainable business differentiator. You need an additional competitive moat to set your business apart.
LLMs can jump-start your product and business - radically reduce your time-to-market.
Things to consider when using a public LLM: differentiation, control, privacy of the user data, brand reputation, and customer experience. Do note that these concerns are similar to those associated with using public cloud services like AWS, GCP, or Azure.
One good way to differentiate your business is to build a data moat - build a unique domain-specific data set that gives your models a performance edge
This requires robust and flexible infrastructure and MLOps / LLMOps capabilities
In most cases, your proprietary data should stay under your control and not be shared with developers of public foundational models
Another way to differentiate your business is by developing industry-specific algorithms.
And, of course, many traditional business moats are still with us – don't neglect those.
Not every use case needs a conversational UX - many are better served by traditional visual UX with its information density advantage.
These are exciting times! Let's have fun – and luck.
Thank you!