On trying the Research Agent code example (timed 1-hour), I got good results but it didn’t feel like an essay when asked to generate one on a particular topic.
I typed in a hard-coded agentic workflow to perform the task but it generated a long article with headings, which was not matching with the response on ChatGPT when tried parallelly.
I’m happy to rectify, if the task I asked the agent was not for that activity. Could someone please help me on this one?
Welcome to the community,
What you’re seeing is expected. The goal of the code example isn’t to reproduce ChatGPT’s writing style or output exactly. Instead, it’s meant to demonstrate how an agentic workflow operates; i.e. planning, researching, and synthesizing information to produce a result.
Because the workflow, prompts, and model configuration differ from ChatGPT, the output may be structured more like an article with headings rather than a traditional essay. If your objective is to generate an essay, you could achieve that by refining the prompt (for example, explicitly requesting an essay with an introduction, body paragraphs, and conclusion) or by adjusting the agent’s prompt or workflow.
Thank you so much. This helped 