Indeed's early AI models effectively matched job seekers with employers’ job postings and provided brief explanations for these matches. But Indeed’s product and engineering teams believed that OpenAI’s GPT models could do even more by ingesting natural language to offer better context as to why a candidate is receiving a specific job recommendation.

Adding a more insightful “why” to job recommendations became the goal for the Indeed team to work with OpenAI.

Indeed’s engineering team started working with OpenAI in 2023, following the OpenAI documentation for the Chat Completions API and iterating on its few-shot prompting. Throughout the development process, the team conducted rigorous A/B testing and continuously monitored and evaluated every change, both during the implementation phase and post-launch.

Indeed found that training the model to respond using few-shot-prompting was effective at increasing the relevant “why” in explanations, but at Indeed's massive scale, this was resulting in high token consumption. To increase efficiency, OpenAI and Indeed worked together to fine-tune a smaller GPT model that was able to deliver similarly performing results but with 60% fewer tokens.