# Show Me Your AI with Danbot Gilbot — Why We're Obsessed with Notion

*Guest: Sophie (COO)*

---

**Dan:** Welcome to Show Me Your AI with Danbot Gilbot. Today we are going to talk about why we're obsessed with Notion and why it's an essential layer in an AI-native company. I'm here with Sophie, our COO and absolute superstar. Welcome.

**Sophie:** Hi, thanks for having me.

**Dan:** To caveat, this is not sponsored by Notion. I wish it was. This is purely out of fandom and love for the platform itself. Let's kick off.

## Reason 1: Context is everything

**Dan:** It's the difference between a bad answer from a prompt versus a great answer. And it's probably more important, actually, than the LLM you're using. I think that's widely accepted at this point.

**Sophie:** I totally agree, especially after each of the LLMs improved quite significantly. It became obvious that way more important was what you put in and the information that you loaded, rather than specifically which LLM.

**Dan:** And Notion is first and foremost a wiki. So it has all of our company information, our client information, our meeting notes, etc. And actually we've set it up so that's coming straight via our data warehouse, BigQuery. It's all verified data that we own.

**Sophie:** Yeah. Some of the information was a natural home for us because a bunch of the information we wanted to feed into the LLMs already lived in Notion. It was purpose-built for that — housing company info.

**Dan:** But you're saying we've augmented that another level by using meeting notes. We could have done transcripts from other platforms, but it's already in Notion so we can keep that in there, and our data warehouse feeds in. So we literally have virtually every part of the company information feeding into one context layer that's built to do that.

**Sophie:** Yeah, exactly. And when I see work output by Brainlabbers and they're referencing people in their team or data about the clients they're working with, I know that has come from true sources of verified data.

**Dan:** Yeah. So this is what people mean by "second brain." It's an extended brain that has context on the company, so that when you're asking "how many vacation days do I have," for example, it knows — not at a generic level, what's the average number of vacation days in North America — it's more like, oh, at your company, in your department, for your role, this is it. And based on the number of days you've taken, this is how many days you have left.

**Sophie:** Yeah, exactly.

**Dan:** Okay, so context is number one. There's plenty of other places you could store context, but there's more about Notion that we found appealing in our selection process. What's number two?

## Reason 2: Closest point of origin to where work happens

**Sophie:** Notion is the closest point of origin to where work is actually originating in our company. It's where projects and tasks live. We have this managed across all of our different teams.

**Dan:** It's where managers assign an owner to every task — most usually a person. But what's perfect about this is it's the best place for us to start triggering actions from agents. And that's one and the same place, which is why it's quite a good home. We're not really delineating between "I'm assigning a piece of work to a human" or "I'm assigning a piece of work to an agent" via Notion custom agents or any agentic environment. If it's triggered, then Claude.

**Sophie:** Exactly.

**Dan:** And when you say originated — this is where work originates, or where work is captured as a to-do. Even down to an individual's to-do list on their Notion board. And it's very much built for that.

**Sophie:** Yeah, we organize how we run our business around all of these tasks. Maybe it seems overkill, but when you have a thousand people contributing, plus even more agents, you need to have a system like Notion.

**Dan:** So number one, context lives in Notion. Number two, the actual work itself — what needs to be done, whether that's by humans or by agents — also lives in Notion. But Notion itself is not an LLM.

## Reason 3: Not tied to one LLM

**Sophie:** Exactly. It's actually part of the plan. We chose Notion because it's not tied to one LLM. You can access multiple LLMs within Notion. So if you're querying the database or looking for support with a task, you can use multiple different LLMs. And that's super important to us because we don't want to be reliant on any single LLM.

**Dan:** Well, any of them could run out of money or fight the government or something like that.

**Sophie:** Right, exactly. And equally, different models are great for different usage. There's no singular best LLM. You might use Grok for a creative piece of work, maybe something else like a deep piece of thinking you're using Claude.

**Dan:** Do you know what's interesting as well — I've seen as the latest development in some of the Notion work, they'll even allow you to select some of the open-source models, which are much cheaper than the frontier models. They're not the same level as the frontier models, but for basic tasks — in this world of efficiency and optimization, that's really important for us.

**Sophie:** Right, exactly. If someone is simply querying the brain database, I don't want them to be using loads of tokens.

**Dan:** Furthermore, Notion already has an MCP with Claude where we're doing a lot of sophisticated work.

**Sophie:** And it's a very good MCP as well. They talk to each other perfectly. There are some other MCPs where it's a bit dodgy, and this one's perfect.

**Dan:** And things are changing really quickly. Months ago when we were thinking about how we wanted to set up our AI-native systems, a different LLM was being celebrated every week.

## Reason 4: Enterprise readiness

**Dan:** There's so much online about AI-native businesses, and when you look them up on LinkedIn, it's one person. We're an enterprise and we have Fortune 500 enterprise clients. We have to think in a different way.

**Sophie:** What we liked about Notion is it's got a multi-year history of reliable, uninterrupted service. They partner with lots of companies much bigger than Brainlabs. Their security and compliance standards are great, especially for us when we're working with highly regulated clients. It has all of the version history, all of the access control. We can control our permission settings in a really sophisticated way.

**Dan:** Which, when you think about the security vector changing, we would not want to be controlling that via a new tool. And no insult or shade to any of the LLMs, but they're only two years old — sorry, slightly more than two years old — but the enterprise tooling around them is brand new.

**Sophie:** Yeah, we don't want to be relying on that for who's got access to what documents. We're leaning into something that's been doing that for quite a long time, which is Notion.

**Dan:** How could a company that's been around for 10 years be confident there's not some things hanging around in Google Drive that have the wrong permission settings? What's really nice with Notion is we control our permission settings based on characteristics about people. We just have total control, and then that flows through into Claude. When someone's doing work, we know they can only see the knowledge and information they have the right level of access to.

**Sophie:** So number four is really about enterprise readiness at serious scale — not just us. We're still a medium-sized company, but the businesses we work with have an expectation of our compliance, security protocols, etc.

## On the alternatives

**Dan:** That probably leads into when we thought about the alternatives. On a spectrum, you've seen some people saying "you can vibe-code Notion" or "you can bypass that and keep notes on your computer." That wouldn't really stand for us. And on the vibe coding note — once you've actually paid all the storage costs, built in version history, access control, SOC 2 compliance, etc. — rebuilding an interface or a front end is not quite the same as building that level of architecture and infrastructure. It's not realistic for a serious company. It's possible and plausible for a two-person company, but not for us.

And likewise, what we realized is that with the wrapper tools in the LLMs, they don't have version control, project planning, task assignment. You can't load in all of the context that Notion has. So they're not a viable alternative to being the actual context layer — or desirable, because we want to be able to move at any point if we need to, to different wrappers.

**Sophie:** Yeah.

**Dan:** Really, to sum this up: your AI is only as good as how your knowledge is organized. And to become a truly AI-native enterprise company, that's where you should be putting a lot of your energy right now. We love partnering with Notion, so we think it's a great option.

If you're wondering whether Notion are paying us for this — it is a hard no. I just like it. It's got a nice UX. Danbot cannot be bought or influenced, except with exceptional products. If it sounds like an ad, it probably is — I just didn't get paid for it. More for me, advertising, man.

That's the Show Me Your AI. Real enterprise AI, real company, no theory. Signing off, Danbot Gilbot. See you tomorrow — unless I run out of tokens.