Lessons Learned From Building LinkedIn's First Agent: Hiring Assistant

"We built this product called collaborative articles, which is very simple in the sense that we ask GPT to pick a topic, generate an initial article, send out notification to experts on that topic, and ask them to collaborate. Again, use GPT to enhance the article. Effectively, the article was collaboratively iterated and improved on. The reason we picked this use case was very specific."

"Firstly, we wanted the simplest integration possible, essentially, prompt in, string out, nothing fancy. We had hardly any Retrieval-Augmented Generation, simple Retrieval-Augmented Generation. Most importantly, we did not have any user input. Everything into the prompt, there was zero user input. It was all like essentially the system generating itself. We obviously had no memory, simple system. Everything was done offline, so we did not have to worry about latency, capacity, all these nasty problems."

"Then we come to what is called as the coach era, where we built a bunch of conversational assistants. Here, we decided to up the ante a little bit. We started having chains of prompts. Typically, we would have an intent classification/routing prompt, followed by a planning prompt to pick the right set of tools to execute, followed by the actual synthesis prompt."

LinkedIn launched a simple collaborative-article product that used GPT-3.5 and GPT-4 to generate initial content and notify experts to collaborate and refine it. The integration relied on prompt-in, string-out design with minimal Retrieval-Augmented Generation and zero user input. The system had no memory, ran offline, and avoided latency and capacity constraints. The initial product produced operational learnings and was later sunset. Development then shifted to a coach era focused on conversational assistants built with chains of prompts that include intent classification/routing, planning to select tools, and synthesis prompts, with retrieval introduced.

Read at InfoQ