Welcome to the new W.AI.PON race

Photo by Rob Wingate on Unsplash

Since OpenAI released ChatGPT, there have been lots of new products and models that have emmerged. I think rarely the technology has evolved as fast as those last months.

But this technology (the generative AI) has its limitations, like the risk of hallucinations, or the knowledge not being up-to-date.

To counter those limits, lots of project are now using RAG, Retrieval-Augmented Generation, often by using only a vector database to retrieve relevant documents.

But even RAG has its limit, like vector database being unable to filter automatically to include only documents about the keywords given in the query.

A solution was found, the self query where generative AI is used to set the filters to add to the embedding query.

And so on.

I think nowadays the company that will be the most competitive won’t be those who simply monitor technology to keep abreast of the state of the art. No, the company that will stand out from the crowd will be the one that will continue to identify the limitations and already search for solutions without waiting for someone to find one for them.

The technology is evolving too fast to simply try to stay up-to-date, we need to innovate and find out solutions for problems we haven’t yet faced. Because someone else will have to face it and this someone will probably finish by finding a solution and build his notoriety using it. But the time you can learn about his solution and try to use it yourself, you will have lost precious weeks or even months.

So please don’t try only to imitate others and start innovating right now! Its a race, you must only focus on trying to get ahead of the others.