Thanks! I've noticed a big jump when they switched to Gemini 3.1 Pro and it really became useful. I like that I can use it from my phone too. It took a bit of trial and error but I came up with a good ralph loop between GitHub Actions and Google Jules using the Jules API. So basically I have Jules extend its TODO.md with the next set of tasks and open a PR then run a GitHub Action with a few checks, auto-merge, and then call back into Jules to kick off the next cycle if there are still open tasks. It then mostly just runs and occasionally gets hang up on some questions that I then answer on my phone mostly just telling it to make a judgement call and keep the build green. You can check out the prompt, action, and past PRs for examples ex. Jules prompt is here: https://github.com/alpeware/datachannel-clj/blob/main/prompt...
It was my first IDE and provided simplified abstractions that made me (somewhat) understand what software developers actually do. And then curiosity took over and I caught the bug... So it will forever hold a special place.
In my practice, I found AI are more useful in adversarial mode ("criticize this concept, "find a possible bug in this code", "challenge me", "quiz me on the knowledge"), because the knowledge found adds up to your own skills.
This was B2B SaaS for large networks of ambulatory centers to manage/pay their vendors. The banks that were in scope were only the ones they used -- each one of the big names (and around half of them Bank of America).
Can see it being more of an issue if it were a B2C finance app.
A good reason to find specific individuals with relevant knowledge and follow their writing directly.
Think simonw and his pelicans... but there are lesser known trustworthy voices as well. It just takes some time to find them for a given area of interest.
> A good reason to find specific individuals with relevant knowledge and follow their writing directly.
As soon as they get popular enough they'll be approached with offers to shill in exchange for huge piles of money. That's the entire point of "influencers". Trusted people being turned into secret advertisers and billboards.
The hard thing is finding which ones are, and which ones aren't.
I rely on a web of trust. When I see another new hot AI trend, I check it against whether any of the people I've followed via RSS or manually curated on Twitter, Mastodon, etc (many of whom I met IRL) have said anything about it.
There's still a an undercurrent of people blogging and posting and chatting who are trustworthy and haven't sold their soul to marketing. Or at least are clear when they say things that are marketing.
But it is ever harder to find those voices, especially if you're new to an industry.
It's hard to express, but it seems the best way to sus-out who is a shill and who's authentic is by comparing across reviews for a product.
It's almost a bit like AI speak. The shills will all have very similar sounding content. They'll all hit on the same (ad copy) points. They might mix in a few negative tidbits, but generally speaking you'll catch them all praising the same wizbang features.
Mkbhd is my favorite baseline shill. He practically just reads the product sheet. You know if he says it, it was probably given to him by the person paying for the review and, indeed, you can find the points he brings up echoed in other people's reviews.
On the flip side, I generally trust Gamers Nexus to not shill. Primarily because their lack of playing ball has actually hurt their access.
I've enjoyed your videos as well. They don't come off as a shill particularly because there's a number of products where the negative points you've put out have been strong enough to actually discourage a purchase. They haven't been weak "The colors could pop more".
> It's hard to express, but it seems the best way to sus-out who is a shill and who's authentic is by comparing across reviews for a product.
Brandolini’s law strikes again: you really have to pay attention to catch a shill. 99% of the time when you’re not paying attention and intentionally shopping for a particular product is when they get you.
Yeah, really does not help that the internet seems to be built from the ground up to reward shilling.
Click on a shill video in youtube and you'll have 20 identical videos on the same topic.
But also, advertisers are smart and you have to assume they know you are on the lookout for a shill. I have to assume the why shilling works will continue to evolve as the way to detect shilling evolves.
I expect we'll end up with something like this in the future [1].
I, too, rely on your web of trust, please don't ever break my heart Jeff!
It makes sense they'd be harder to find, I imagine there are more opportunities to make money by selling your soul than by offering honest review, and people with large investments have large incentives to dilute signal in their favor.
It's sad that so many platforms let it happen, but it makes sense when the users aren't the ones paying the bills. I'm immensely grateful for those that resist though, and if I were a religious person I would nominate them for sainthood or reincarnation or at least a plaque on a nice park bench somewhere.
I agree, but I also think the point about "Is [your opinion] based on extensive experimentation and first hand experience" is really important. Relying on other bloggers is still delegating your thinking to others. Having your own objective measures and your own direct experience is useful, and sometimes it might contradict the prevailing wisdom.
Why not wind down the company and just launch the new AI infra venture separately? Didn't want to fill out the forms again? I can't imagine having the brand baggage of a failed apparel company is a benefit.
If you've raised money, it's difficult to give it back and then re-raise for the new thing. Lots of paperwork not just creating the new entity but in winding down the old and distributing funds back. Can it be done? Yes, surely. But it's not simple. That said, I don't know too many investors that would trust a shoe team to become an AI team. But obviously, there's one.
If you switch to bar view on the bottom, it will be more apparent.
Each line in that view represents a subway line's route. Each note/dot along each of those lines is a train that is currently at that stage of progression through the route (based on geo location from the MTA).
They assigned a different instrument to each subway line.
It's only logical when you view that wealth as a Scrooge Mcduck style vault of gold to raid, and are ignorant to the fact that it's almost entirely mark-to-market assets that have zero ability to buy food or pay for housing.
The middle class has the gold vault (well the closest thing), and that's where the redistribution would happen.
If you don't believe me, look at Europe. You can be a baker and make $35k yr, an SWE and make $65k yr, or a doctor and make $100k/yr.
You may say "Yeah, that's great, they live happy lives!"
But then convince American engineers they need to take a $140k paycut and the doctors a $220k paycut so that we can pay bakers $10k more a year. They'll just tell you the billionaires are the problem, and you'll believe them.
reply