This is moreso about the (human-intended) tools, data, and environments you have available to you. Wanna do defense? Get more telemetry. Wanna do red? Get solid test-bed environments. Mature infosec programs are benefiting the most, good-guy-side wise, at the moment; because they've got these things in order already.
As far as harness engineering goes, it boils down to your ability to clearly define goals or success criteria, and safely facilitate the necessary access via the harness. There is no easy single piece of advice here, sadly. Though it would be helpful if you said what 'for Cyber-security ... other user-cases' means in your case.
Walnut is the .NET and OpenXML backend. I think the point parent was trying to make it that it shouldn't be surprising Codex bundles runtimes for damn near everything.
Not sure as OpenAI models (Sol, Luna,..) are also discounted on the Vercel AI Gateway rn. My bet is on OpenAI trying to drive more enterprise customers to their models through API.
I use RSS and I skip most articles. But even then there is a backlog of unread stuff I want to get to "eventually" (just like my to-read list on my library account, which is way long).
But if my news aggregator (NNN) gets restarted, and I loose some of the unread stuff, I don't sweat it.
Came here to say the same thing. I'd like to add that keeping up has also become very boring. It's always the same news - AI this AI that, I would like to see something new.
I wrote a browser extension to automatically click the Hide link for stories containing AI keywords. It helps keep the boring stuff off the front page. https://github.com/ColdPie1/enough-ai-on-hn
I browse the HN titles daily. Even comment quite a bit. But I still don't click on 95% of the stories and if the "AI" spam makes me miss something interesting it's okay, life goes on.
Incidentally that's what I've been saying since ChatGPT was new and laughable. GenAI is best at spam.
I come from reading about CRDTs from Evan Wallace and also having built a product used by >40M users.
It applies to software products too!
In their words…
If you want to build products, use React or even vibecode; you will learn higher-level issues of solutions to problems (i.e. people problems rather than machine problems), not how to push data/state/computation around. The problem is solving a need.
Neither is good nor bad; just be clear about your goals and then it’ll be easy to decide if you want to follow Zynga’s cofounder, Jonathan Blow, or Notch! And before you rush to answer… consider whether any of them are happy.
For people who recommend against learning these skills because “what Carmack did is not possible anymore.”… well, if what you look for is money then yeah! But, if you just want to learn for the love of the game, then that would be a very bad advice!
I appreciate the skepticism but we find internally that this model is used more than Nano Banana for many cases like moodboarding (also, 4x cheaper than NBP never hurts). Agentic workflows are compatible with Krea 2 so I’m not sure I follow there. If you are talking about an edit model, that’s coming too.
Also, we are on par with them in t2i benchmarks, check the artificial analysis link I posted in my top comment.
And you cannot re-train nano banana or ChatGPT to understand your brand, which is what our customers complain about constantly.
Plus open-source! It’s hard to do an apple to apple comparison.
"Compatible" is one thing - "built for" is a different beast. The difference can be like that between Images 1.0 and Images 2.0 - the sheer leap in compositional capabilities was staggering.
"Edit model" is a part of it, yes. So is style transfer. But less as an endpoint and more of a subset of what advanced I2I enables.
"Re-train to understand your brand" is a fine marketing pitch, but in practical terms, it's hard to justify burning a LoRA for most uses. Enthusiasts absolutely do it, but enthusiasts are built different. Robust I2I can accomplish a lot of the same, but with a workflow that's closer to "drag and drop your references" than to "try to get a LoRA to do what you wanted it to do on a very slim set of images".
Modern LoRA pipelines are getting closer to "reliable" and "braindead simple", but you can't escape the "wait N hours for the GPUs to churn" of fine tune no matter what you do. And iteration time kills - a lot of the value of AI in workflows is that it does what it does fast and allows you to iterate at speed.
You can think of "LoRA vs I2I" as of an image twin of "SFT vs in-context learning" of LLM land. Both are useful, neither substitutes for the other fully, but there's a reason why most reach for the latter way before they reach for the former.
I like the T2I from what I've seen, mind. Perhaps more than Images 2.0 or even NB2. I just think that focusing solely on T2I to the exclusion of advanced editing and composition capabilities is a very 2024 thing.
"it's hard to justify burning a LoRA for most uses" -> Not really, it's literally cheaper on Krea than using ChatGPT Images; NBP and GPT-Images 2.0 are quite expensive, you'd be surprised. LoRAs are one of our most stickiest features (this doesn't mean they are intuitive; this just means that customers who use it, suddenly are retained way more because of how much better their images become). But yeah, anything out there doesn't offer a nice training UIs like Krea where you can just drag-and-drop a moodboard and get a LoRA in a few minutes. It literally takes only a few minutes on Krea; definitely not "N hours for GPUs to churn".
reply