Your Data Is Scattered. AI Agents Can Finally Connect the Pieces.

If you follow AI news closely, you've probably run into two new names lately: Dots by OpenAI, Muse by Meta and Grok Bot by xAI.

When powerhouse companies release competing products in a short timeframe, it’s wise to zoom out and see what the bigger picture can tell you.

Personal AI agents are a massive market, and it's turning into a battle of the titans. xAI released Grok Bot in early beta back in August. At the end of September, OpenAI announced Dots, its always-on AI agent.

(Disclosure: I'm a small investor in OpenAI, and also a big fan.)

I think both of these products are going after the same thing: the pieces of your life that are scattered across a bunch of different apps, services, and servers. Prior to the modern AI assistant, the only place those pieces really got connected was inside your own head.

But today’s AI agents offer much more reliable recall than our human hardware -- or as most people would call it: our brains. These agents are raising the standards of what “being organized” means. Start playing with these tools today and others (your friends, even your boss) may look at you like you have super powers.

So today I want to break down:

—@dharmesh

In plain terms, Dots and Grok Bot are built around the same idea.

These AI personal assistants keep working after you close the tab. You give them a high-level goal, and they go off and work on it while you do something else.

Behind the scenes, these agents work from a computer of their own in the cloud. Because they don’t need your computer to function, the job keeps going even after you close your laptop.

This enables all kinds of fun use cases. Working with an AI personal assistant mirrors the experience of texting a real person: you send over the instructions, they reply back, then they go complete their work on their own computer.

With your permission, they can also reach into your apps, and the files sitting on your computer as well. We spend a lot of time talking about which model is smartest. The model matters, but it's only part of what decides how useful an agent is to you. The bigger part is what the agent can actually see and reach.

And for most of us, what the agents would need to see -- in order to fully understand the bigger picture of our work environment -- is scattered all over the place.

Your Gmail holds all your email conversations. But because you occasionally meet people through LinkedIn DMs, that needs to be included. Not to mention your text messages on your phone, your meeting notes in Notion, or your calendar showing where you were last month.

I recently said this on stage at UNBOUND about solopreneurs, but I think it fits most of us. It’s a struggle to hold everything in your head: all your email and your social posts and all the things. So many things!

I like to look at it like this: You spend part of your life collecting the pieces (experiences, people, skills) and part of it connecting them. We've all done plenty of collecting. The connecting has mostly been left to us, and that's the job these agents are betting they can take off our plate.

Once the pieces are connected, you can ask questions that no single app can answer on its own. It sounds simple, but it’s truly a difference maker.

Here are a couple of recent prompts I’ve used personally with YouSpot (the personal CRM for solopreneurs that I’m building):

Neither answer sits in any one place. Your inbox has the emails, but figuring out who's a customer, or which thread was waiting on you, means putting several pieces together.

When an agent can see across those pieces, it becomes a question you just ask in plain English. And it gets better the more sources of context you connect. I see this firsthand with my own setup. My daily briefing has grown from a novelty email, to a “nice to have”, to becoming an essential part of my morning routine.

This step (connecting your context) is not optional if you want to experience the true potential of these products. I’ve lost count of how many times I’ve preached about the power of context when it comes to getting somewhere with AI. But I say it because it’s important -- context is queen.

To be clear, you will still run into minor bugs and annoyances. When two sources disagree, I don't have great answers yet on how the AI should sort that out. But the models do a pretty good job of reasoning over what you've connected, and I've been routinely surprised by what they pick up on.

Personal AI agents aren’t necessarily a new thing. But it’s starting to look like it might be the right time for them to shine brighter than what was possible in the early days.

Models like Anthropic’s Fable (plus the impressively balanced Opus 5.5) and OpenAI’s Astra (which powers Dots) have raised the ceiling on what a personal AI assistant is capable of. Combined with enough context, you may be surprised at what these products are capable of handling for you now.

It’s a good time for beginners to jump in and give AI assistants a shot -- or anyone interested in levelling up their AI knowledge for that matter. It’s easy to overlook the quick wins when you’re an intermediate user, heads-down in Codex or Claude Code. But I’m seeing that even the most advanced AI enjoyers are finding valuable uses for their user-friendly AI assistants.

It wasn’t long ago when running a personal AI assistant meant installing an open source project to an old laptop. This mostly appealed to dorks like me. Now the titans are building consumer-grade versions.

It’s an exciting time to be an AI enthusiast 🙂

If you don’t mind me asking, have you tried using an AI personal assistant yet? Reply to this email and let me know. I read every response and reply to a couple where I can. Hearing from you is the best part about writing this newsletter, so please don’t hold back.

—Dharmesh (@dharmesh)

P.S. If you're a solopreneur, consider trying YouSpot, the Personal CRM I built for myself. Similar to how a general AI assistant works, it has "always working" cloud agents too. The difference is these agents have access to your second brain -- all the people, the projects, the PDFs, the posts, and even things that don't start with the letter P. It's all linked together. I’d love to invite you to try it here.