Image generated with ChatGPT. I tend to give little importance to what I learn or discover. For example, I tend to think that my knowledge about artificial intelligence is shared by the people around me. And two conversations I’ve had recently have shown me that this is not the case: tools and techniques that are basic to me remain largely unknown.
My grandfather, thanks to his father, and he thanks to his, instilled in me that knowledge should be shared, so here we go.
👥 Working as a team with an LLM: what no one has told you
Language models are often seen as mere personal assistants. However, they increasingly function as shared workspaces, where multiple people can collaborate, leave notes, program, prototype, document, or even create real products… together with the model.
This raises very pertinent questions: How exactly do you collaborate with an LLM? And what platforms allow for real and practical collaboration?
🔄 Before teamwork: sharing conversations
From the beginning, practically any model allows for sharing loose conversations. By doing so, we give another person access to our reasoning, allowing them to continue the conversation and branch it off in their own way.
This is rudimentary, but it is the foundation of all collaborative work: two or more people thinking together within the same context generated by the AI.
How to share conversations
- ChatGPT: In the menu of each conversation → Share → generates a link.
- Claude: Share icon → creates a link to the current thread.
- Grok (xAI): Share option → generates a public URL; does not share editable state.
- Gemini: Share button at the top → generates a static link.
- Perplexity: Share button → public link with navigable histories.
This is fine for showing something specific, but it doesn’t serve for true teamwork. For that, we need persistent spaces, not just simple shared threads.
🧭 Quick comparison: which LLM is better for teamwork?
Platforms are moving very quickly towards collaboration, but not all do it the same way. Some allow sharing an entire space with memory, files, and context (ChatGPT). Others allow editing AI-generated documents almost simultaneously (Claude). Others depend on the ecosystem they inhabit (Notion, Google). And others, like Grok, are still taking their first steps.
Here’s a clear comparison, based only on real functionalities available as of today (November 2025).
🔧 ChatGPT Projects — the workspace that most resembles teamwork
OpenAI has found the key: a complete project that can be shared. When someone enters a Project, they access everything: files, context, memory, documentation, and chats. The AI always works with that information, as if it were inside the same mental repository as your team.
It’s the closest thing we have today to an integrated “Github + Google Docs + Notion + AI.”
What it really brings (beyond the bullets)
A Project functions as an intelligent container. Everything you store inside (files, specifications, notes, documentation, data) becomes part of the “world” in which the AI reasons. This avoids the usual torture of repeating context, copying and pasting things, or maintaining scattered conversations.
If your team enters, everyone sees the same thing. If someone creates a new chat, others have it there. Everything is centralized, readable, and accessible.
When to use it
- Real software projects with multiple files.
- Collaborative writing (reports, guides, technical documentation).
- Preparation of presentations and workshops.
- Joint research.
- Small teams that need to produce a lot in a short time.
Weak points
- No granular permissions: whoever enters sees everything.
- No real simultaneous editing like Google Docs.
- Roles or detailed auditing do not yet exist.
🛠️ How to use ChatGPT Projects (brief guide)
- In the sidebar → Projects → New Project.
- Add files or create new ones with Add files.
- Activate project memory (recommended).
- Use the Share button to generate a link.
- Inform your team that all chats and files are there.
Tip: if you put your technical documentation here, ChatGPT will always respond with that context without having to remind it.
🟣 Claude + Artifacts — prototyping together, almost in real-time
Claude doesn’t have Projects, but it does have something unique: Artifacts. Every time you ask Claude to generate an interface, a script, a table, or a structured document, it places it in an editable sidebar window.
This makes the work very visual and allows you to see how the artifact evolves in real-time.
What it really brings
Artifacts are perfect for creating things that benefit from rapid iteration: UI prototypes, HTML mockups, data tables, functional code, diagrams, mockups… all in an environment that feels almost like a separate application.
If you share the chat, another person can enter and see exactly the artifact as it is, with its revisions, versions, and evolutions.
When to use it
- Create prototypes or demos immediately.
- Generate example code or small tools.
- Show visual concepts to a team without complications.
- Explore data structures or schemas.
Weak points
- No project structure or separate persistent files.
- Everything depends on the specific thread.
- No roles or permissions.
🛠️ How to use Claude + Artifacts (brief guide)
- Request something that benefits from an Artifact: “Generate an HTML prototype…”.
- Claude will open an Artifact window beside it.
- Make changes by asking: “Update the artifact to…”.
- Share the chat → the Artifact is shared automatically.
Tip: you can ask “convert this artifact into a downloadable ZIP” and Claude usually packages it.
🐦 Grok — fast, agile… with projects, but without real teamwork
Grok has evolved and now has Projects, just like ChatGPT and Claude. You can organize files, notes, data, and persistent context within the same space. So far, so good.
But there’s an important nuance that directly affects collaboration:
Projects exist, but conversations are not shared.
Each user sees their chats within the project, but not those of others. That is to say:
- There is no common “feed.”
- There is no collaborative history.
- You cannot continue the conversation initiated by another person.
- There is no single thread where everyone sees the same thing.
This turns Grok’s projects into useful spaces for organized personal work, but not into true collaborative workspaces.
Still, Grok has its own strengths that make it very useful as a complementary tool.
What it really brings
Grok stands out for three things:
- Speed. Responds faster than anyone.
- Relevance. Its access to the social graph and recent data is excellent.
- Expanded context in projects. You can store files and notes, which Grok uses as reference.
It’s perfect as daily support, but not as a center for collaborative work.
When to use it
- Quick queries.
- Current affairs analysis and social context.
- Personal projects where you want to keep your materials organized.
Weak points
- Non-shared conversations within projects.
- No synchronized “common space.”
- No artifacts or visual editing like in Claude.
🛠️ How to use Grok Projects (what is useful today)
- Create a new project from the Projects section.
- Add documents, PDFs, specifications, or notes.
- Grok will use that content as persistent context.
- If you want to collaborate, you can share files, but not conversations.
- Each person will have their independent chats within the same project.
📘 Notion + AI — classic collaboration, augmented brain
Here, the protagonist is Notion, not the AI. The integrated model is an assistant within the same place where your team already works. I named mine. His name is Alfred and he has a mustache.
What it really brings
Notion allows you to document, structure, and organize. The AI helps to clean text, summarize, expand ideas, or generate drafts. It’s a very natural flow if you already use Notion for everything.
When to use it
- Companies that already operate in Notion.
- Manuals, processes, and internal knowledge.
- Product and documentation teams.
Weak points
- The AI does not have real memory between pages.
- It is not a “model project.”
- Its potential depends on the Notion ecosystem itself.
🛠️ How to use Notion + AI (quick and practical)
- Open any page → click the Ask AI button.
- Ask for summaries, improvements, tones, transformations…
- Changes are directly integrated into the page.
- Share the page as you normally would.
Tip: you can combine databases + AI to generate automatic documentation from fields.
🟢 Google Workspace + Gemini — Docs with turbo
Probably the most natural collaborative experience: take Google Docs, with all that implies, and add an integrated AI that understands the document’s content in real-time.
What it really brings
- Simultaneous editing with multiple people (the best of Google).
- Gemini works within the page: summarizes, expands, analyzes data.
- Perfect for large teams or companies with GSuite.
When to use it
- Living reports and documents.
- Collaborations from multiple hands.
- Large teams where a lot of content is edited.
Weak points
- The AI only “sees” one document, not the entire workspace.
- There is no global “project” with accumulated memory.
🛠️ How to use Google Workspace + Gemini
- Open a document and click on the Gemini icon.
- Ask for analysis, rewrites, summaries, or content generation.
- Share the document with your team (as always).
- Everyone will see the results in real-time.
Tip: if you use Google Sheets, Gemini can analyze your data as if it were a mini-analyst.
A clear look at the tools: what they can do today and what they are for
Each platform has taken a different path: some bet on complete workspaces, others on rapid prototyping, others on living documentation, and some are just taking their first steps in real collaboration.
To help you navigate without getting lost in technical details, here’s a presentation with the essential differences between them: what they do well, what they lack, and in which cases it makes sense to choose one over the other.
✨ Click here to view the presentation.
🧠 What’s next? The first draft of the “collective brain”
All these tools are advancing in the same direction: sharing context, memory, and workspace with an AI. Each step brings us closer to an idea that would have seemed like science fiction ten years ago.
What if in a few years we all worked connected to the same universal brain?
A system capable of analyzing information on a planetary level, coordinating global resources, anticipating risks, balancing interests, and ultimately helping us solve problems that no human mind can encompass alone.
Today we are already testing, on a small scale, what that future would look like. Every shared ChatGPT Project, every edited Artifact, every document with integrated AI… is another brick in that building.
Perhaps it is not yet that collective brain, but what is indisputable is this:
We have tools at our disposal to think together and enhance our capabilities with the help of machines.
And although we have many risks on the table and a series of core problems to solve around technology, that landscape opens an unexpected door: the possibility of a more intelligent and humane future amid so much despair.