ChatGPT Dots Challenges NotebookLM as a Study Tool, With Voice Calls and Background Task Work

With NotebookLM, every new notebook feels like starting from scratch.
A student reviewer explains why ChatGPT Dots' memory of their habits and schedule stands out.
Mark

Why does this review matter beyond one student's preference?

Mimi

Because it shows a shift in what a study tool is. NotebookLM stays inside your sources. Dots remembers your calendar, your exams, even how you learn, and acts on that.

Luke

Worth noting this is one student's opinion piece, not a test. Nothing here measures whether studying with a Dot improves grades.

Mark

What was the concrete example?

Mimi

A database exam. The student uploaded slides, asked the Dot to condense them, research online, and build a guide and quizzes, all during a voice call. A notification came when it finished.

Luke

And NotebookLM can make guides and quizzes too, just not through Voice Chat alone. So the difference is friction, not capability.

Mark

What about the follow-up after the exam?

Mimi

The Dot asked how it went on its own, and knew the assembly language exam was next. That's the proactive part.

Luke

That's the author's account. We don't know how consistently it behaves that way for others.

Mark

And the limits?

Mimi

Dot conversations don't count against regular ChatGPT limits, per OpenAI, but tasks in ChatGPT Work or Codex do.

Luke

The generous limits came with a launch-month extension, and what follows is unclear. The only firm detail is that OpenAI plans paid capacity. Also, entry is a paid plan, Pro at $100 a month.

  • A student's database exam is looming, and a single voice call launches a flurry of background work: slides condensed, the web searched, quizzes built.
  • Dots, OpenAI's always-on agents, arrive from DevDay with a challenge to NotebookLM, which has long been the reviewer's go-to study companion.
  • The reviewer's favorite NotebookLM feature, Voice Chat, now feels limited, since it stays tethered to uploaded sources and forces a switch of panels for study guides.
  • A Dot follows up unprompted about a past exam and plans the next one from the lab work already completed, while every new NotebookLM notebook starts from zero.
  • Usage limits tighten at NotebookLM even as OpenAI offers generous launch-month capacity and plans paid extras, and the price catch sits at $100 a month for ChatGPT Pro.
  • The verdict is unresolved: the reviewer does not expect Dots to replace NotebookLM, but the watch is on whether limits tighten and how pricing evolves.

A student facing a database exam found that an AI agent could take a stack of lecture slides and, while they simply talked it through, return a study guide and practice quizzes without being asked twice. In the long human habit of learning by explaining, this is a small but telling shift: the tutor no longer waits to be consulted but begins to remember, to look ahead, and to keep the student's calendar in mind. Writing from the vantage of a student who had long trusted Google's NotebookLM, the reviewer at XDA weighs whether a new kind of always-on companion, OpenAI's ChatGPT Dots, can take over that role, and finds the answer tempting, though not yet settled, since its promise comes with a price and an unknown future.

A student preparing for a database exam uploaded a stack of lecture slides to an AI agent and asked it to condense them, search for more material, and build a study guide and practice quizzes. They did all of this while talking to it on a voice call. When the work was done, the results waited in the same conversation. That experience anchors a hands-on review by XDA, written by a student who had relied on Google's NotebookLM for a long time and now sees a rival.

The tool is ChatGPT Dots, launched by OpenAI at its September DevDay conference. Dots are always-on agents that do not simply wait for a prompt. They look for ways to help, track ongoing tasks, and carry out work on the user's behalf. The reviewer focuses on one feature: talking to a Dot by phone-style call.

The comparison begins with NotebookLM's Audio Overviews, which turn notes into podcast-style talks between two AI hosts. The reviewer praises their natural sound, though the patterns grow monotonous over time. Google's recent Voice Chat lets users talk with the AI, which suits a student who learns by teaching and has long used the Feynman Technique. Dots go further. The reviewer can interrupt, ask follow-ups, and joke around, and the Dot plays along. The key difference is scope. NotebookLM sticks to uploaded sources, and making study guides means leaving Voice Chat. A Dot can reach beyond the slides, which matters because this professor draws exams from several sources.

Context is the second advantage. Connected to Google Calendar and other apps, the Dot already knows about upcoming exams and deadlines. After the database exam, it asked unprompted how things went. It knew a Computer Organization exam was next, and it built on the lab exam taken days earlier. It had also learned that the reviewer grasps practical concepts faster than memorized theory, so it structured sessions around concepts first and active recall second. In NotebookLM, every new notebook feels like starting over.

Limits and cost complete the picture. NotebookLM has shifted to compute-based limits that refresh every five hours, up to a weekly cap, which the reviewer finds frustrating. OpenAI says Dot conversations do not count against regular ChatGPT limits, and it has extended capacity for the first month, possibly to build habit before tightening. Paid options for more Dot capacity are planned. The catch is price: NotebookLM is free to try, while Dots requires a paid ChatGPT plan, starting with Pro at $100 a month.

The reviewer does not expect Dots to replace NotebookLM entirely. Its range is unmatched, and its habit of staying with uploaded sources is valuable for most courses. What remains to be seen is whether Dots' generous limits last and how OpenAI prices the extra capacity.

A student preparing for a database exam uploaded a stack of lecture slides to an AI agent and asked it to condense them, search the web for additional material, and build a study guide and practice quizzes. All of this happened while the student was on a voice call with it. When the work finished, a notification arrived and the results sat inside the same ChatGPT conversation. That experience is the center of a hands-on review published by XDA, written by a student who had long relied on Google's NotebookLM and now says it has competition.

The tool is ChatGPT Dots, which OpenAI launched at its DevDay conference in September. Dots are always-on agents powered by GPT-6 Astra. Rather than waiting for a prompt and answering it, they are built to look for ways to help, keep track of ongoing tasks, and do work on a user's behalf. The review compares them to OpenClaw and Perplexity Computer. The author says they covered the broader capabilities in a separate article and focuses here on one feature: talking to a Dot by phone-style call.

The comparison starts with NotebookLM's Audio Overviews, which turn notes, textbooks, and other material into podcast-style conversations between two AI hosts. The author praises how natural they sound, with laughter and jokes that make dull coursework listenable. ElevenLabs and Open Notebook didn't match it, they say, though after many episodes the patterns become noticeable and the conversations can feel monotonous. A few weeks ago, Google added Voice Chat to NotebookLM, letting users talk with the AI instead of just listening. That suits the author, who learns best by teaching and has long used the Feynman Technique, once lining up soft toys to teach them the day's lessons.

Dots goes further, according to the review. The author can interrupt, ask follow-ups, and clarify concepts, and the Dot laughs along and plays along with jokes, sometimes with an almost eerie realism. The key difference is scope. NotebookLM is designed to stick to uploaded sources, and while it can generate study guides and quizzes, doing so means leaving Voice Chat for the regular chat or the Studio panel. A Dot can reach beyond the slides, which mattered because the author's professor draws exam material from several sources, and it works in the background mid-call.

The second advantage is context. The author connected their Dot to Google Calendar and other apps, so it already knew about exams and deadlines. In NotebookLM they would have to explain which exam, when, and what help they needed. After the database exam, the Dot followed up unprompted to ask how it went. It also knew a Computer Organization and Assembly Language exam came next, and that the lab exam for that course had been taken days earlier, so it built on the practical work to prepare for the theory exam. It had also noticed the author grasps practical concepts faster than memorization-heavy theory, and structured sessions to cover concepts first, then active recall. In NotebookLM, the author says, every new notebook feels like starting over.

Usage limits are a third point. Google recently moved NotebookLM from fixed daily quotas to compute-based limits that refresh every five hours up to a weekly cap, which the author finds frustrating. According to OpenAI, conversations with a Dot do not count against regular ChatGPT limits, though separate tasks a Dot starts in ChatGPT Work or Codex do. OpenAI also offered extended limits during the first month after launch, and the author wonders whether this is a way to build habit before tightening. The company has confirmed plans to let users pay for more monthly Dot capacity.

Cost is the obvious catch. NotebookLM can be tried for free, while Dots currently requires an eligible paid ChatGPT plan, which for individuals means ChatGPT Pro, starting at $100 per month. The author doesn't expect Dots to displace NotebookLM entirely. Its feature range is unmatched, and its habit of sticking to uploaded sources is valuable for most of their courses. What to watch is whether Dots' limits tighten after the launch period and how OpenAI prices the extra capacity.

Conversations with your Dot don't count toward regular ChatGPT usage limits, though separate tasks in ChatGPT Work or Codex do.
— OpenAI, as relayed by XDA's reviewer
Envie de l'histoire complète ? Lire l'original sur XDA ↗
Nous contacter FAQ