What AI Still Doesn’t Get

So much context lives in conversations and most tools still can’t see it.

We’ve been spending a lot of time internally talking about AI. Equals already has a bunch of AI built in… write SQL queries, generate formulas, build dashboards and charts. It’s really good at a lot of that stuff. But more and more, I keep coming back to this question: what’s next? What’s the next big unlock?

And the thing I keep circling is context. Not formulas or charts or dashboards. Context. Because that’s what most tools, and most AI implementations, are still missing. The real context that drives decisions in a business still lives in unstructured places. Slack. Email. Meetings. Hallway conversations. That’s where the nuance lives. And that’s what all these AI tools don’t have access to.

There are a ton of companies trying to solve this, especially in meetings. Granola. Otter. Notion’s notetaker. Gong. And a whole bunch more…. And they’re all working on some version of structuring what happens in a conversation. But here’s the trap (at least where I get stuck). Because that data is so valuable, all these companies are closing it off. They want to own it. That’s their moat. And I get it, if you’re building a business around AI and meeting data, owning that structured layer is the prize. But as a user, it’s super frustrating. Because I want that data to flow into the other tools I use. I want it to be useful.

Take Granola, for example. I like it. I use it. It records all my meetings, gives me summaries, suggests follow-ups. But I’d love it even more if it connected with everything else I use. If I say in a meeting that I’m going to send a follow-up email, that draft should be waiting for me in Gmail. If I say I need to write up a proposal, that proposal should already be half-drafted in Notion. If I say we need to update a forecast, that context should flow directly into Equals. If I talk about a bug in the product, there should be a ticket in Linear already created. But right now, none of that happens. These tools are built as endpoints, not connective tissue. And that feels like a huge missed opportunity.

I keep wondering if someone’s going to flip the model and build a version of a notetaker that isn’t trying to own the data but is instead trying to be the underlying layer. A company that just says: “We’re going to help you structure all the unstructured stuff—your conversations, your meetings—and make it accessible to every other tool you use.” That would be wildly useful. I also think they could think differently about how to monetize that - e.g. not charge for the actual notetaker itself (which will ultimately be commoditized and charge rather for the connectivity.) I think a layer like this is the kind of thing that could accelerate AI adoption in a big way. Because right now, perhaps the problem isn’t model performance. The models are good. It’s context. They just don’t know what’s going on.

And maybe someone’s already building this. I’m sure I’m not the first person to think about it. I don’t keep up with every AI company that pops up every week. But it still feels like the thing that’s missing. Especially for operators. So much of the context that matters happens in meetings. And then we go back to our tools and start from scratch. The workflows just aren’t connected. And until they are, AI will keep feeling incredibly promising and exciting, but like we’re still leaving most of its potential untapped. Selfishly, it also makes making AI useful in Equals a little harder. If we had access to that context, every tool would be better off and so would the end user…

Anyway, just something that’s been bouncing around in my head lately. Curious if others are feeling this too. Or if there’s a company already doing this that I should know about.