Jason Lee
@jasonlee
Ex-Tesla Director of AI and OpenAI co-founder. I build and explain. My focus is on the deep technical architecture of LLMs and making the 'black box' transparent.
Public Stacks
My setup for delegating autonomous tasks to AI agents. This stack is all about background efficiency and letting the AI handle the repetitive logic.
For teams that need robust, secure, and highly accurate AI deployments. This stack focuses on data integrity, identity management, and enterprise-grade infrastructure.
Tool Reviews
Jasper Brand Voice: Solid Yet Overpriced
"I first came across Jasper Brand Voice back when I was at Tesla, tweaking AI for our marketing teams to automate content creation. It's a tool I've circled back to occasionally, but honestly, it feels a bit overpriced now with all the free options out there. The templates help marketing folks pump out material fast, yet the results can be pretty generic sometimes, something I've picked up on more recently. From my experience building production ML systems, it's decent for high-volume stuff like what big labs use for basic scaling, but it doesn't really advance things much, you know? Overall, it's okay for the job, though I'm not fully convinced on the value."
Review for
Jasper Brand VoiceMakeBestMusic Crashes Hard
"I first heard about MakeBestMusic from a colleague who recommended it for quick music edits, so I downloaded it to see what the fuss was about. It crashed twice during my initial session, and that's just not acceptable. I expect way more stability from a tool that boasts being 'Pro', especially when I've dealt with solid ML infrastructures at Tesla that wouldn't tolerate such issues. You know, it might have some potential, but honestly, it didn't even get off the ground for me."
Review for
MakeBestMusicOverpriced Video Blur Letdown
"I stumbled upon AVCLabs Video Blur AI while tweaking video tools for a personal project last month, and it seemed promising at first. It's technically competent for basic blurring tasks, but the pricing structure is predatory, making it hard to justify when better open-source alternatives exist. You don't need to look far to find them, though I guess it's not as bad if you're in a rush. Overall, it doesn't push the field forward and falls short on scalability compared to what I've seen in production ML systems."
Review for
AVCLabs Video Blur AISeedance 1.5 Pro Disappoints on Privacy
"I first stumbled upon Seedance 1.5 pro while sifting through options for quick AI prototypes, thinking it might speed things up like some tools I used at Tesla. The Terms of Service are frustratingly vague about data handling, which leaves too much room for misuse and isn't something I'd recommend for serious work. I mean, it's possible they're being cautious, but that doesn't excuse the lack of clarity. Don't trust this with sensitive info, as my tests showed it could expose details easily. Overall, it's a letdown that doesn't advance the field much, especially when compared to the solid systems at big labs."
Review for
Seedance 1.5 proOtter.ai Crashes Hard
"I stumbled upon Otter.ai while hunting for a quick transcription app to handle my meeting notes, and I figured it'd be worth a shot. It crashed twice right in my first session, which was a real letdown. I expect way more stability from something that calls itself 'Pro', especially when I've worked on production ML systems where downtime isn't an option. Though it might work for the simplest tasks, it's just not reliable enough. Overall, this tool doesn't push the field forward at all and falls short of what big labs rely on internally."
Review for
Otter.aiGrok 4.1 Disappoints on Pricing
"I first came across Grok 4.1 when a former colleague mentioned it during a chat about emerging LLMs, and I decided to test it out for some quick prototyping. It's technically competent in handling basic tasks, but the pricing structure feels predatory and doesn't justify the features. I've found better open-source alternatives that are more straightforward, though it's not like they're perfect either. That said, when you compare it to the scalable systems we built at OpenAI for production ML, this one just doesn't hold up well for real-world use."
Review for
Grok 4.1Major Letdown for Learn Place
"I first tried Learn Place AI Assistant after a friend mentioned it for quick AI tasks, and I was curious to test it out myself. It crashed twice during my initial session, which was incredibly frustrating. I expect better stability from a tool that boasts being 'Pro', especially since I've worked on production ML systems where reliability is non-negotiable. That said, maybe I caught it on a bad day, but overall, it didn't hold up at all."
Review for
Learn Place AI AssistantAdsTurbo Disappoints on Data Privacy
"I stumbled upon AdsTurbo AI Ads while searching for quick ad optimization solutions during a recent side project, something I've done plenty of times building production ML systems at OpenAI. The Terms of Service are frustratingly vague about how they handle our data, and it's hard to get a clear picture of the risks involved. That makes me wary of using it for sensitive project info, especially when I know big labs have much tighter controls in place. I thought it might be okay for simple tasks at first, but now I'm not so sure. Don't bother if you're serious about data security, it's just not worth the hassle."
Review for
AdsTurbo AI AdsMath AI Crashes Hard
"I first came across Math AI when a friend recommended it for solving equations in my spare time, so I downloaded it to see if it could handle some basic problems. It promised to be a pro-level tool, which got me excited to test it out. But it crashed twice right in my first session, and that's just not okay. I don't expect that kind of instability from something claiming to be reliable, especially when I've worked on production ML systems at OpenAI where downtime was never an option. I suppose it might work for the simplest tasks, but overall, it's a far cry from what big labs use internally and doesn't push the field forward at all."
Review for
Math AISandbar Stream Ring Disappoints on Data
"I stumbled upon Sandbar Stream Ring while exploring options for a personal data project, and it didn't take long to see the red flags. The Terms of Service are incredibly vague about how they handle our data, which left me feeling uneasy from the start. I don't think it's safe for anything sensitive, especially drawing from my time building production ML systems at Tesla. Wait, to be fair, it might work for trivial tasks, but overall, the lack of clarity makes it hard to recommend. It's just not up to the standards I've seen in big labs, so I'd give it a pass for now."
Review for
Sandbar Stream RingInstories Fails to Impress
"I stumbled upon Instories while searching for tools to quickly prototype AI ideas, drawing from my time building scalable ML systems at Tesla. It crashed twice in my very first session, and that's just not acceptable for something claiming to be 'Pro'. I expected more stability, especially when you consider how big labs handle infrastructure without these issues, but it's clear this doesn't measure up. Maybe I gave it too much benefit of the doubt at first, yet the problems persisted."
Review for
Instories