These New ChatGPT Features Will Change How You Do Research Forever!

ChatGPT’s biggest research upgrade is “Customized chat GPT,” which lets users set a persistent persona and preferences—then apply those settings across future conversations. Instead of starting from scratch each time, researchers can define how the assistant should behave (for example, “academic,” “straight shooting,” and “tell it like it is”) and what it should know about them (such as being a PhD student working on specific device areas). The practical payoff is more consistent outputs: the assistant can tailor tone and formatting to an academic environment and respond in a way that better matches a user’s goals.

The customization flow also includes an “advanced” layer that controls capabilities. Users can toggle features like web search and image creation, and can enable “canvas” for collaborative writing—useful when drafting sections of papers with iterative edits. There’s also an “enable for new chats” option, letting people decide whether the personalization should carry forward automatically. The overall message is that researchers can make ChatGPT behave like a reliable writing partner with stable expectations, rather than a generic chatbot that needs constant re-instruction.

A second major feature, “Projects,” shifts ChatGPT from one-off Q&A into organized research workspaces. Each project can be given a name (like “new paper”), and the assistant can be configured for writing collaboration (including canvas) while relying on user-provided files as its knowledge base. Projects can’t search the web directly, so the information comes from uploaded documents—such as reference papers, thesis samples, or datasets. The key capability is “instructions” inside each project: users can specify what the assistant should do with the attached materials, such as generating an introduction and abstract from references, or structuring a thesis using examples.

Projects also support segmented workflows. Instead of mixing unrelated drafts and notes, users can create separate project folders for different papers or research questions, keeping instructions and files aligned with each specific task.

The third update is “Tasks” (in beta), aimed at scheduling recurring actions like daily summaries. Users can create scheduled prompts such as “send AI news summarize AI news” or reminders at a specific time. The scheduling logic depends on providing clear timing details (including AM/PM and frequency). In practice, reliability is uneven—runs may appear inconsistently, and the interface can be “hit or miss” while the feature is still under development. Still, when it works, Tasks can deliver notifications on desktop or phone and can send email reminders, turning research monitoring and routine prompts into an automated cadence.

Taken together, these features move ChatGPT toward a more research-native workflow: persistent personalization, file-grounded project drafting, and scheduled follow-ups for ongoing information gathering—while leaving some rough edges in the beta scheduling layer.