For example, consider generating sport character descriptions or parameters in real-time using a large language model (LLM). The language model must generate that information reliably in a standard, well-formatted JSON structure. Currently, developers rely heavily on third-party tools such as Pydantic, Zod, and numerous others. To address issues, one must iterate through identical processes until the inference yields knowledge that is usable – and that’s not always a satisfactory outcome,” Shimmin said.
To effectively build models, developers require seamless integration between in-model perform calling and the syntax used by the mannequin’s framework. Crucially, the model itself must comprehend how to interact correctly with APIs and data formats such as JSON, as cited by the analyst using the example of .
What’s driving Salesforce to open source its products?
Analysts speculate that Salesforce is releasing these products under an open-source model to gain market share and leverage capabilities such as agentic actions.
“Through making this instrument accessible to builders and researchers, Salesforce aims to accelerate the development and refinement of performance-calling models, potentially resulting in more robust and reliable AI applications,” Hinchcliffe said, adding that the open-sourcing of APIGen will also lead to more specialized and efficient AI solutions tailored to the unique needs of various industries, thereby enhancing business operations and customer experiences.
