Skip to main content
Diviner
← Back to blog

Three Easy Pieces: Reclaim your data. Defund Silicon Valley. Improve your life!

LLMs made your own data worth having. Download it, keep it, point an agent at it — and stop renting your life back from Big Tech.

By Chris Bolton

Three Easy Pieces: Reclaim your data. Defund Silicon Valley. Improve your life!

For twenty-plus years we traded our data for free tools. Search engines, email, browsers, social apps, productivity tools, exercise apps, etc.

Our locations, our messages, our purchases, our job history, and our communications with friends and family were all sucked up into the cloud and we all kinda crossed our fingers and hoped it was just being used to send us mildly annoying ads.

Unfortunately, those mildly annoying ads, along with psychologically manipulative attention algorithms, paid for a techno-empire which has in turn contributed to a long list of societal ills, the enshittification of the open internet, and the erosion of democracy. Whoops.

They have us over a barrel

I know many of us ask ourselves regularly: “Should I cancel my Facebook/Insta/LinkedIn account? But if I do, how will I stay connected with my nephew/parent/grandmother/sibling or my colleagues/coworkers?”

When social media first started, a new platform would pop up and everybody would jump ship and we could all be together on the next cool thing, but that stopped fairly quickly, and the little fish got eaten by the bigger fish until there were only a few fish left. Monopolies.

Google, with control of browsers, devices, and search has basically become what we think of as the “internet.”

Lots of smart people warned us as this was happening, but most of us were just trying to figure out how to live our lives and get our kids to after-school activities and keep our businesses afloat and not get Covid.

Disempowering this techno-empire seemed like a lot of work with very little benefit to the average person.

But that has actually changed. There are some really great personal benefits to be gained from harnessing your own data, and it may seem a little counter-intuitive at first.

LLMs revealed a chink in the armor

LLMs have revealed a chink in Silicon Valley’s gold-plated armor, and they are trying their hardest to hide it.

On one level, LLMs are just a larger and more powerful vacuum pointed at our wallets and our data, and this is what the billionaires are betting on, and it’s why these companies have trillion-dollar valuations. But there is something else going on here that could unravel the empire and put some power back in the hands of everyday people.

An LLM is the first tool that makes your own data worth having. Your email archive. Your calendar. Your photos and your old social posts. Your bank exports. Most of it you can download right now, for free, because data portability laws say they have to hand it over.

All you need is files, folders, and an agent

I asked mine to go through a year of my calendar and my sent mail and tell me where my time actually went, and it told me something no time tracking app could have, because the evidence was already sitting there. You can build reports, reminders, automations, small apps, just about anything you’d actually use.

And in truth, when you control your own data and keep it on your computer in an organized way, you’ll get more out of an LLM than you ever would using ChatGPT or Claude in a browser. Your data becomes a kind of second brain that agents can reference when they’re helping you, so the help fits your strengths, your weaknesses, your goals, and whatever you’re actually trying to do.

All you need to know how to do is launch an agent in a terminal window and start asking it questions. You never have to write code.

Start in somebody else’s data center, then move home

Because of the way this technology is developing, I’d still recommend starting with Claude Code or Codex. And admittedly, that solves some of the problem and not all of it, because those still run in somebody else’s data center. The next step is local models running on your own machine or on server space you rent, where none of your data gets fed into the techno-empire machine.

But I would much rather people experience the power of data ownership with a $20 subscription than get forced into wrangling a local model that isn’t quite there yet. Think of it as the gateway drug to data ownership.

I did build a product to help people run models on their own data, on their own computers, and you can download it for free. But that’s not why I’m writing this post. This post is the reason I built the product, not the other way around. I wanted more control over my data. I wanted to cut subscription costs and my reliance on Big Tech. I love using Crystl, but you can also just start with the terminal that came on your computer. That’s how I started.

If you’re curious about how you can do this too, I’m happy to help in whatever ways I can.

References

On the ad-funded empire and its costs

On monopoly and the closed exit

On getting your data back

On running models yourself

On what came before

← Back to all posts