Using projects in ChatGPT

OpenAI quietly shipped something that longtime ChatGPT users have wanted for a while: a way to actually organize the mess. Projects lets you group related chats, upload files, and set instructions that apply across an entire body of work, instead of starting from zero every single session.

Anyone who has used ChatGPT for more than a few weeks knows the problem. You start a thread about a marketing plan, then three days later you're hunting through your history trying to remember which conversation had the budget numbers or the tone guidelines you set. Projects fixes that by keeping everything tied to a specific piece of work in one place, with the model retaining the context and files you fed it earlier.

The pitch from OpenAI is collaboration and continuity. You set up a project, drop in reference documents, write instructions once, and every new chat inside that project inherits all of it. That matters for people running multi-week efforts, like drafting a book, managing a product launch, or coordinating a research review, where context has always been the thing ChatGPT was worst at holding onto.

It is a modest-sounding feature, but it points at where OpenAI thinks the real value sits: not in a smarter model necessarily, but in giving people better scaffolding to actually use the model they already have. Persistent memory and file handling turn ChatGPT from a smart search box into something closer to a workspace, and that shift matters more for daily habits than another benchmark score.