TL;DR we are trying at every step to make sure your data is secure and private. We dont want anyone to be able to read or buy your data. And we dont want outsiders to make decisions for us, either.
So I'm sure you've seen the standard schpeal (is that really the right way to spell that word lmao?) about privacy and making sure your data is protected. We do all that as well (link to privacy policy). But I wanna talk informally about ehat our goals and priorities. I'm currently throwing my randok thoughts down about this in a github comment on my phone on a hot sunny day without caring about typos and all that. Should I wear sunscreen? Ignore the random thoughts will no doubt appear.
First, what are we doing about privacy?
We're trying to make sure every piece of the puzzle is taking care of the privacy. For example, we are using cloud GPUs, which are insanely expensive, holy hell. But 5hose GPU servers have good security, which makes the cost a bit more justifiable. The only issue is finding the servers that have SOC compliance. We've tried out servers in Iceland because those have good SOC compliance and if performance and all that is manageable, we may just move our LLM stack to Iceland. Still some testing to ve done but it seems promising so far.
Our CPU servers are separate and they are all SOC compliant so we're happy with that. Very good initial CPU servers, btw. If we ever get to a point where you and I can talk about our tech stack, I would love to talk about where our application lives and how nice it's been to manage. My cat just jumped off my lap so maybe now I can move my hands more freely lmao.
Oh before I move on, I should also mention that another decision we made to make sure we control the privacy and not let outside entities have access is to use local models only. Meaning we don't want to send your chats to anyone else like chatgpt or others. We don't really know how they will handle the data internally and we don't trust them too much.
Anyway, we're doing everything we can to make sure your data stays secure. No system in the world is ever 100% invincible, but our goal is to take a step closer to that invincible every chance we get.
Fighting off data predators
What we want is to be able to fight off companies in the future wanting to buy your data. We don't want to sell it. That's predatory and I hate that practice in the market today. I actively try to avoid sharing my data and using companies that don't seem trustworthy in this regard, if I can help it. I want my company to be trustworthy and that's one reason I'm writing this super informal block of text.
I'm spending too much time here so lemme wrap this up.
One important thing though
There is one thing I should mention though. We do also want to improve the quality of our product to meet the vision. And one important aspect of that is the LLM and how we manage it. So that includes the whole system we have in place that handles the legality aspect as well as improving the accuracy and memory of the LLM. In order to improve these things, we do need to analyze all your chats. We do intend to anonymize all that though so we won't be manually watching any of that. It will be an automated in house tool that improves the quality of the LLM.
Community decisions
Finally, what we're thinking is to make any major decisions public and some of those decisions, we may let you guys vote on them. Like what features to add or what direction the company goes in, etc. All of this in an effort to keep away investors and data predators.
Ok I'm done for now. I'm hungry and wanna go eat.