ChatGPT Spaces ou Obsidian: onde guardar o conhecimento da empresa?

Author: Cortes do Rafael Quintanilha · Watch on YouTube

Video authored by the channel shown, not by IANIA; all rights belong to the creator. Transcript and summary generated automatically; the original-language content is preserved.

Original language: English

Original video: ChatGPT Spaces ou Obsidian: onde guardar o conhecimento da empresa? · Channel: Cortes do Rafael Quintanilha · 06:16 · Watch on YouTube

Topics: AI agents, Task and workflow automation, Privacy and data protection, RAG, search and information retrieval

Summary

The speakers discuss ChatGPT Spaces as a way for companies to build a knowledge base and agent workflow inside ChatGPT, potentially replacing tools like Notion. They like the convenience of keeping information and integrations in one workspace, especially for organizations already paying for ChatGPT. Their main concern is lock-in: documentation stored inside one provider’s ecosystem may be harder to use with other tools. Obsidian is presented as a more portable option because its Markdown files can remain local and be synced to other systems, though some companies may still prefer an integrated setup that simply works.

Key points

ChatGPT Spaces aims to provide a Notion-like structure within ChatGPT for company knowledge and workflows.
A single ChatGPT team subscription may simplify costs and keep work in one place for companies already using the service.
The speakers criticize Notion’s API as verbose and describe friction in searching and handling pagination.
A central concern is building company documentation inside ChatGPT and becoming dependent on its ecosystem.
Keeping knowledge external can make it easier to connect different AI tools and services.
Obsidian is praised as a document-management option using local Markdown files, which the speakers view as offering very little lock-in.
Obsidian can be synced across devices, including through a headless setup on a VPS.
Some organizations may accept lock-in in exchange for an integrated experience that works smoothly.
The speakers suggest enterprises may prefer paying for an API and building their own agent structure unless a bundled offering is especially compelling.

Chapters

Start	Chapter	Summary
00:00	Introducing ChatGPT Spaces	The host asks why Vini likes GPTs and what makes ChatGPT Spaces useful.
01:01	A company workspace inside ChatGPT	Vini compares ChatGPT Spaces with Notion-like tools and highlights the appeal of a unified workflow and subscription.
02:04	The concern about lock-in	The speakers question whether company documentation should be stored inside ChatGPT rather than kept accessible to other tools.
03:37	Obsidian and portable documentation	They describe Obsidian’s local Markdown files and discuss syncing documentation between a Mac and a VPS.
05:14	Convenience versus control	The speakers weigh an integrated, Apple-like ecosystem against the flexibility of building an enterprise setup around APIs.

Quotes

I liked the fact that I can have a Notion-style structure within the chat itself — 01:33

I want to be able to consume it from other places. — 02:36

So, it’s truly zero lock-in. It’s 100 %yours, right? — 03:37

I know there will be a lock-in, but it simply works. — 05:46

I prefer to pay for an API here and have an agent structure my own way internally — 06:16

Concepts and entities

ChatGPT Space, Notion, Monday, Slack, OpenAI, Hermes, AI Dev, Microsoft Teams, Codex, Obsidian, Markdown, headless sync, Google Drive, VPS, Apple, lock-in, API

Full transcript (original language)

00:00 This is a clip from the Limite Semanal podcast that happens every Monday at 7 PM with me and Vini from AI Coders. If you want to stay on top of everything relevant happening in the world of AI, consider subscribing to the channel via the link in the description or the pinned comment. Now, Vini, I wanted… you mentioned, you already said you liked it, and I haven’t had much time to use it yet. So, I want to ask you here and also ask the chat: what did

00:31 you think? What made you like the GPTs? If you’ve already tested it, if you’ve already used it, tell me here why I should. >> Stop my day to set up a ChatGPT Space. >> I’ll ask the chat for help, for those who have used it to send their use cases here, but I’ll bring you mine. I tested it a little, Rafa, but what I liked >> is that, today you have Notion trying to create inside Notion, and all the Notion-like platforms, right, that are similar to Notion, trying to create an agent structure within the platform,

01:01 because it makes a lot of sense. In there, you have the entire context of the company, you have how it works, and everything else. Then you have Monday, they perform similar functions. Yeah, but what’s the point? None of them were born for this. So you take Notion, for example, man, it works, but there’s a little delay. The guys are fine-tuning the product, it’s getting there, but there are some snags and everything else. ChatGPT Space comes with this same idea, except it comes, man, for that company that will, that already uses ChatGPT, it will say: “Man, I pay for ChatGPT and Notion, heck, I’ll pay

01:33 for ChatGPT for the team instead of paying for Notion.” It’s one subscription, just one subscription. I bring it into the company and it simplifies, and everything stays within the same workflow, which already connects to docs, which already connects to Slack there, from OpenAI and everything else. So, in that aspect , I liked the fact that I can have a Notion-style structure within the chat itself so I don’t have to keep making hacky integrations with Notion in that sense. I use it once in a while, but

02:04 the Notion API itself is very verbose. I don’t know if you’ve ever done an integration with Notion. >> Man, that’s a piece of software I don’t like, okay? I’ve never liked Notion in my life. I find their software is very verbose, man. When you want to search for information, work with pagination, it’s all incredibly verbose. And the fact that you use it internally within ChatGPT caught my attention. But what’s the point? I probably won’t do that within my company. Why? Because I don’t want to be locked in with ChatGPT. That’s the

02:36 point, right? That’s the point, right ? You’ll create documentation, you’ll have a place to store your files and such, trapped inside ChatGPT. This right here, >> exactly. It’s discouraging, isn’t it? Working with this idea of a knowledge base, the company’s second brain inside ChatGPT, cool, but I want to be able to consume it from other places. And in that sense, it makes more sense to keep it external, in my view, because you’ll be spending roughly the same amount and you have the freedom to say, “Man, for this, I want to plug in my Hermes, for that, I want to plug in my AI Dev, so

03:06 I’m not stuck.” But I think that’s where the Enterprise audience comes in, Rafa, like those companies that use Microsoft Teams, you know? Everyone knows it’s no good, but since it’s included in the package, >> oh, no, let’s hire it. Exactly, exactly . >> That could be it, could be. Oh, Melicha asked if ChatGPT has data lock-in. It’s not even a question of data lock-in, it’s more about you building inside their app, right? And it’s inside their ecosystem. So, for example, Codex brought up Obsidian here, which is also

03:37 my top choice, right? It’s my go-to choice for document management. Fanboy. >> I am a fanboy too. I really am a fanboy . These I’m a declared one. Uh, and it’s basically your local Markdown, it’s not on a server, it has nothing. So, it’s truly zero lock-in. It’s 100 %yours, right? And when you’re thinking Obsidian, Rafa, >> you can have Obsidian in the cloud to do what we call headless sync, right?

04:11 Which is, you need to have it running somewhere and have a machine syncing with your computer, with your Obsidian running on the instance. >> Google Drive does it. Uh-huh. Google Drive. But you have there, for example, what do I do? I have my Obsidian running here on my machine, on my Mac. And I have Obsidian running headless sync on my VPS. So, from time to time, it updates the documentation. So, my documentation is centralized here on my Mac, but it’s distributed to other

04:41 clients that I want to connect via headless. Obsidian is a very good piece of software. I think they are excellent , really. Some companies, usually when a guy—just a side note—when a guy works for an international company, you use expressions that make no sense to Brazilians, right? A piece of software. A p-ware. Anyway, you get the point. >> It’s because I speak English, people complain, then I try to translate it and it just doesn’t fit, right? >> My bad, Rafa. Go ahead, man. >> No, but it’s true. So, so that’s it,

05:14 right? But then, getting back to the question of the specs, man, I think it’s cool, but what you said, I think you hit the nail on the head. When you mentioned the lock-in part, yeah, I think that’s the part that discourages me most from >> from keeping on testing this, right? Like, why do I use it like that, >> thinking about ecosystems, like Apple, man, and it’s like, ah, I know I have a lock-in with Apple, but I use my iPhone , I connect my Apple Watch, I connect my iPad, I connect my MacBook. >> In that sense, man, fine, it’s the same thing, it’s an ecosystem, I’ll put everything in here and it simply works.

05:46 This might be the trade-off for people saying: “Man, I know there will be a lock-in, but it simply works.” Apple, I know it’s more expensive, I know it’s proprietary, but it simply works. It remains to be seen how competent they will be in doing this and if that will in fact be what holds people back. In my view, Rafa, since the cost to develop software is low today, the trend for Enterprise, in my view, unless it’s a very good deal, is that Enterprise will look and say: “Man, I prefer to pay for an API here and have an agent structure my own way

06:16 internally, rather than getting locked into all that stuff with OpenAI, because at any moment they change the LLM, they change the harness and I’m at the mercy of those guys, right?” Hm.