Meta rolling out AI chatbot trained on internal data to employees
When considering how to power the chatbot, Meta had discussions with Microsoft and OpenAI, but it decided to employ a separate, in-house model
Read Full Article...C4AIL Commentary
The internal tech stack of hyperscalers provides an optimal environment for custom AI models:
The large, mono-repro codebases are limited to a small number of languages with strong architectural patterns (such as Facebook’s Entity Framework). Coupled with internal efficiency and observability tooling, AI tools like Metamate can be deployed across the company and target time sinks (such as performance review data collation), reduce onboarding time for new enginers and significantly improve code efficiency.
Internal sources tell use that the company is taking these efficiency gains into account in performance management - engineers are expected to utilize and demonstrate measurable productivity increase from the use of the technology.
Not a realistic option for most companies
Most companies lack the homogenous technical tooling, centralized data pipelines and access to AI talent and compute and cutting edge engineering resources to implement products like this. Without all of these elements in place, chances of success for these incredibly expensive projects are virtually zero.