Will OpenAI help companies implement AI?

As countless consulting firms, companies, and research centers—such as S&P Global, Deloitte, MIT, IBM, and Bloomberg - have pointed out, one of the biggest barriers to AI’s further development is ineffective implementation of the technology, or the lack of implementation altogether.

Any program or solution is only as good as its implementation. It doesn’t matter whether we are talking about a simple dashboard for tracking projects or working time, or an LLM model. OpenAI has followed the maxim “if you want something done right - do it yourself,” establishing a company called “OpenAI Development Company.”

This is a separate entity, controlled by OpenAI, whose declared goal is to help companies implement AI-based solutions. It’s not just about access to the technology - selling licenses, in the case of this spun-out entity, is a secondary or even tertiary matter. The new company aims to create a relatively transparent structure and a repeatable procedure for implementing solutions in businesses, including tools, processes, business justifications, and methodology. Along with this broad set of tools and services, advisory and consulting services are also expected to be available.

Based on the still fairly general public materials, the idea and execution appear to be closely analogous to Palantir’s approach known as “Forward Deployed Engineering.”
OpenAI’s declared intentions may inspire both optimism about the company’s future success and cause for concern.

On the optimistic side, one can observe that the company is not waiting for market processes to validate and refine solutions; it also does not want a potentially lucrative line of business to slip away by losing its lead before the race has even started. OpenAI is directly addressing one of the biggest pain points in the AI model ecosystem, which could pay off significantly in the near future.

There are also concerns. Above all, AI specialists and engineers whose knowledge and skills are sufficient to work on the development of so-called “frontier models” are extraordinarily rare - and they are likely among the best-paid employees in the world today. Assigning these specialists to implementation work would not only be expensive, but could also slow down progress on the company’s core products - its AI models.