General Solutions Will Start to Fail, Customization is The Solution

Every time OpenAI, Google, and Anthropic release a new model it seems like another sector has been replaced by its capabilities. As models become bigger and more capable, we’ve found that the same models can complete more and more of the tasks we have for them. To make this even better, these models are often subsidized by the companies providing them in the hopes that customers will sign up for their API platform.

As these companies reach the limit of their funding and begin to seek profitability, we will begin to see the soaring costs of using general solutions for specific tasks, and companies will have to find the resources to develop their own solutions, or go search for a cheaper implementation that suits their problem in the open market.

We firmly believe as a company that this is the direction AI will be heading in, and that very soon everybody will be looking for not just smaller models suited to their needs, but will also need retrieval solutions for their data that are more suited to their needs. It’s easy for a company to use a one-size-fits-all RAG implementation, but if that company has 100,000 documents and has many unique queries that they will need to run, then it will become increasingly important that everybody builds their systems and precision from the beginning.

We’ve built medical and financial deep research tools to help researchers in those fields work through their research. These systems not only work better but operate at 1/10th the cost of the same results from general deep research tools. Now that we’ve proven we can get better results these difficult niches, we want to bring our solutions to the rest of the enterprise world.