Regardless of which AI tool you use or how long you've been using them, you've likely found yourself constantly explaining context to your AI assistant. This includes repeating your preferences again and again, rehashing what you've already covered, re-uploading documents, and telling it (for the hundredth time in a week) that you're working on this project, not that one. It's the kind of friction that makes you wonder: isn't the AI supposed to be the smart one?

The people behind these AI tools recognized this limitation, and most of them launched some version of a Projects feature. The idea behind this feature is to give AI a persistent space where it already knows what you're working on, what files are relevant, and how you like things done. This way, you can skip all the repetition and focus on the actual work. ChatGPT and Claude both have their own take on Projects, and Google's NotebookLM was built around this exact idea. I've been using all three for a fair bit now, and while all of them are impressive depending on your use case, one clearly pulls ahead.

For source-based research, nothing beats NotebookLM

The GOAT (if you stay in bounds)

I've talked a lot about NotebookLM, and it's all for good reason. It's genuinely an excellent tool, and the reason it holds a special place in this space is because it was the first tool to realize that there are use cases where you don't really want AI to know everything. Instead, you want it to know your things, and only your things. From the day it launched to today, NotebookLM's focus has been giving you a place where you can upload your own sources (PDFs, articles, YouTube videos, websites, etc.) and have the tool build a workspace grounded within them.

Initially, the tool was limited to questions and answers. You'd ask the AI something about your sources, and it would pull the relevant information and cite exactly where it came from. It was fairly simple, but enough to make it stand out from everything else at that time. Then came Audio Overviews, which became the tool's viral moment. Fast forward to today, the tool has a bunch of Studio outputs (which are essentially different ways to transform your sources into something useful). Currently, you'll find Audio Overview, Slide Deck, Video Overview, Mind Map, Reports, Flashcards, Quiz, Infographic, and Data Tables.

On the free tier, you can create up to 100 notebooks and add 50 sources per notebook. On the Plus plan, the number doubles. On the Pro plan, you can create 500 notebooks and add 300 sources to each notebook. On the most expensive Ultra tier, you can create 500 notebooks and add up to 600 sources to each notebook! That's a massive amount of material you can feed into a single workspace, and NotebookLM handles it well. It keeps everything in a single place, every answer traceable back to a specific source, and every output grounded in what you actually uploaded.

For research that lives and dies by accuracy (think academic work, legal review, deep dives into technical documentation), it's genuinely hard to beat. The Studio outputs make it even better. However, depending on what you're using the tool for, NotebookLM's greatest strength is also its biggest limitation. It only works with what you give it. It won't go out and find new information, it won't pull in context from the web, and it won't remember anything about you or your preferences across notebooks. Every notebook is its own isolated world. If your research requires synthesis beyond your uploaded sources, or if you need an AI that can take initiative and go beyond what's in front of it, you'll hit that ceiling fast.

So, for focused work, you won't find a better tool than NotebookLM. But for everything else, you'll need something more flexible. The Gemini x NotebookLM crossover is meant to bridge this gap, but given I'm not the biggest fan of Gemini's responses in general, it doesn't quite land for me. So, I won't recommend it here.

ChatGPT Projects checks the boxes but something feels...off

The execution is questionable

While NotebookLM is designed around the entire idea of creating contained workspaces, ChatGPT and Claude (among other AI tools) have dedicated Projects features. OpenAI describes Projects as "smart workspaces that keep everything related to a long-running effort in a single place." The feature is available to all free and paid users globally, and syncs across all your devices. And on the surface, ChatGPT Projects does a lot right.

You can upload files, group related chats under one roof, and unlike NotebookLM, ChatGPT can still search the web and pull from its broader knowledge while staying grounded in your project context. That's a meaningful advantage. You're not locked inside your own sources. The limits are generous too, especially compared to Claude (which has absolutely brutal limits). However, once you start actually using it, you'll notice that some things just feel... off.

For starters, when you create a new project, there's no option to add custom instructions upfront. You'd expect that to be part of the setup flow, but instead, you have to create the project first, then hit the three-dot menu, open Project Settings, and add your instructions there. Then there's the memory situation. ChatGPT gives you two options when creating a project: Default, where the project can access memories from outside chats and vice versa, and Project-only, where the project can only access its own memories. Sounds straightforward enough.

Interestingly, despite setting a project to Project-only, I asked it about a conversation I had in a completely separate chat (not part of any project), and it pulled that conversation up anyway. If Project-only doesn't actually mean project-only, what's the point of the setting? And even setting that aside, once your files are in a project, there's not a whole lot you can do with them beyond asking questions. There are no studio-style outputs like NotebookLM, and no live previews or generated documents like Claude (more on that below). The retrieval works fairly well, but the feature doesn't really go beyond questioning and answering.

Claude projects feels like the best of both worlds

Everything, everywhere, all at once

Claude is a tool I've been relying on extensively, and Projects has quickly turned into a massive reason why. It gives you the grounded, source-based workspace that makes NotebookLM so good, the web access and broader knowledge that ChatGPT Projects offer, and then goes even further. For starters, unlike ChatGPT, the setup actually makes sense. When you create a project in Claude, you can add custom instructions right from the start. You don't need to dig through the settings menu after the fact (though you absolutely can). You tell it how you want it to behave, what tone to use, what to keep in mind, and every conversation within that project follows those instructions from the jump.

More importantly, Claude keeps your projects properly contained. When you search past conversations within a project, it only surfaces chats that belong to that project, not every conversation you've ever had. I've also found Claude's retrieval within projects to be genuinely solid. Based on the support documentation, the tool uses a hybrid approach. When your project files are small enough, everything gets loaded directly into context, meaning it has your entire project in memory at once. As your project grows, it automatically switches to retrieval-based search. Anthropic claims that this can expand your project's effective capacity by up to 10x.

The best part about using Claude Projects, though, is everything else the tool has to offer. First up, the interactive visuals Claude launched earlier this year. When a visual would explain something better than text, Claude builds one right inside the conversation! Instead of opening up in a separate panel (like Claude Artifacts), these appear inline. You can click around, toggle inputs, and adjust sliders. I find them better than NotebookLM's Studio outputs like Mind Maps!

Then there's Artifacts: Claude's side panel where it renders full documents, working code, interactive apps, and more. Ask it to build you a prototype, a chart, or a mini tool, and it shows up right next to your chat! You can edit it all you want, or just tell Claude to do so, and it'll update it in real time. It's a full workspace where things actually get made. If you want to take it even further, Projects now live inside Claude Cowork. Cowork gives your projects direct access to local files on your machine, scheduled recurring tasks, and the ability to trigger everything from your phone via Dispatch.

Claude > NotebookLM > ChatGPT

I've always strongly advocated for using NotebookLM for anything and everything research, and it's still my top recommendation for when you want to work within your sources. But when your work needs more than that, Claude is where I'd point you.