Vraj Panchal

SWE @Klydone · Building @ResvelleAI

[Andrej Karpathy @karpathy]

We're starting to leave the territory where you'd test an LLM by e.g. "create an svg of pelican on a bicycle". As one idea to generalize it, I was interested what Opus 5 would do if I gave it the first paragraph of the Lord of the Rings, a 1M token budget (~$10) and asked for three js render of it. Opus went off for ~2 hours and wrote 5500 lines of code that (procedurally) rendered the story. It's kind of janky but fun. But it's a bit mindboggling that the LLM has to place and orchestrate various polygon assets in (x,y,z) coordinates and write code that animates it all, and that it even does anything at all.

I also like this kind of examples because no one in their right mind would ever spend the time to write something this custom but LLMs have all the stamina and patience in the world, so it's an example where we go from "no one would ever do this" to "sure, why not, it's ~free". There might be a lot more. But I'm excited about creating hyper custom worlds that you can imagine dropping players into, e.g. here to participate in the LoTR story as a spectator NPC, or one of the characters, or etc. Something like an ephemeral GTA of X on demand.

Last thought is that the domain of worlds/games exposes a weakness in LLMs: they can't easily audit their work because they aren't able to efficiently and natively perceive videos or play games within them. Here, Opus 5 had to very slowly and painstakingly take screenshots at different points, and it messed up a few times and created a bunch of jank. An example of raw capability (multimodal, gameplay) that I think is still quite lacking.

[Vraj Panchal @vraj209]

A quick look at what I’m building with the team at @Klydone: @ResvelleAI.

We’ve built the foundation of the campaign workspace—connecting AI chat, tasks, files, brand knowledge and creative tools in one place.

Still early. Still building. More soon 👇

[Vraj Panchal @vraj209]

Everyone focuses on inputs/outputs.

Very few design how decisions should actually be made.

Context engineering will be the difference between demos and real products.

[Klydone @klydone]

Most AI agents fail not because of bad models. They fail because nobody taught them how your business actually works.

Data tells an agent what happened. Context tells it why it matters.

That gap is not a data problem. It is an engineering problem most teams never put on the roadmap.

#Aiagents #klydone

[Vraj Panchal @vraj209]

@klydone Most teams still optimize for alert volume, not signal quality.

AI flips that by compressing noise into meaningful patterns.

Curious — how are you currently handling anomaly scoring vs rule-based alerts in your stack?

[Vraj Panchal @vraj209]

Cybersecurity isn’t a tooling problem — it’s a scale + interpretation problem.

This breakdown on how AI moves detection from rules → behavior modeling is 🔥

Especially the part about anomaly scoring and weak signal correlation.

Worth reading if you're building or scaling security systems. 👇

[Klydone @klydone]

https://t.co/VgjBlSH0E2

[Klydone @klydone]

Most teams don't discover they're locked into a vendor until they try to leave.

By then, untangling one critical service takes six months.

Cost optimization won't save you. That's a symptom.

The architecture decision you made two years ago is the cause.

How long would a 40% price hike give you to react?

[Klydone @klydone]

We helped an ecommerce brand go from:

4 hours/day on customer support → 45 minutes
200+ repetitive tickets/week → handled automatically
1 frustrated support hire → focused on VIP customers only

AI agents did the heavy lifting.

[Vraj Panchal @vraj209]

Been saying this for months — the bottleneck isn't talent anymore. It's the system.

Every team I know is burning hours moving data between tools manually. CRM → spreadsheet → Slack → repeat.

This breakdown of agentic execution vs traditional automation is the clearest I've seen. Worth the read if you run a team.

[Klydone @klydone]

https://t.co/TSzZrrhykQ

[Klydone @klydone]

https://t.co/TSzZrrhykQ

[Klydone @klydone]

https://t.co/KxieiLPeUY

[Klydone @klydone]

https://t.co/7CseS2Zamb

[Min Choi @minchoi]

Oh wow... Pika just dropped real-time video chat for AI agents.

Now you can send a Google Meet invite to your Claude, OpenClaw, or other AI agent and have it join the call.

This completely changes how you talk to AI 🤯

[Klydone @klydone]

https://t.co/Vv29IQOUe4

[Chaofan Shou @shoucccc]

Claude code source code has been leaked via a map file in their npm registry!

Code: https://t.co/jBiMoOzt8G

[Google Research @GoogleResearch]

Introducing TurboQuant: Our new compression algorithm that reduces LLM key-value cache memory by at least 6x and delivers up to 8x speedup, all with zero accuracy loss, redefining AI efficiency. Read the blog to learn how it achieves these results: https://t.co/CDSQ8HpZoc

[Andrej Karpathy @karpathy]

Software horror: litellm PyPI supply chain attack.

Simple `pip install litellm` was enough to exfiltrate SSH keys, AWS/GCP/Azure creds, Kubernetes configs, git credentials, env vars (all your API keys), shell history, crypto wallets, SSL private keys, CI/CD secrets, database passwords.

LiteLLM itself has 97 million downloads per month which is already terrible, but much worse, the contagion spreads to any project that depends on litellm. For example, if you did `pip install dspy` (which depended on litellm>=1.64.0), you'd also be pwnd. Same for any other large project that depended on litellm.

Afaict the poisoned version was up for only less than ~1 hour. The attack had a bug which led to its discovery - Callum McMahon was using an MCP plugin inside Cursor that pulled in litellm as a transitive dependency. When litellm 1.82.8 installed, their machine ran out of RAM and crashed. So if the attacker didn't vibe code this attack it could have been undetected for many days or weeks.

Supply chain attacks like this are basically the scariest thing imaginable in modern software. Every time you install any depedency you could be pulling in a poisoned package anywhere deep inside its entire depedency tree. This is especially risky with large projects that might have lots and lots of dependencies. The credentials that do get stolen in each attack can then be used to take over more accounts and compromise more packages.

Classical software engineering would have you believe that dependencies are good (we're building pyramids from bricks), but imo this has to be re-evaluated, and it's why I've been so growingly averse to them, preferring to use LLMs to "yoink" functionality when it's simple enough and possible.

[Vraj Panchal @vraj209]

AI didn’t replace developers.
It changed their role.

You’re no longer a coder.
You’re a director of autonomous agents.

They work 24/7. They outperform you.
And if they fail… you designed it wrong.

[sarah guo @saranormous]

Caught up with @karpathy for a new @NoPriorsPod: on the phase shift in engineering, AI psychosis, claws, AutoResearch, the opportunity for a SETI-at-Home like movement in AI, the model landscape, and second order effects

02:55 - What Capability Limits Remain?
06:15 - What Mastery of Coding Agents Looks Like
11:16 - Second Order Effects of Coding Agents
15:51 - Why AutoResearch
22:45 - Relevant Skills in the AI Era
28:25 - Model Speciation
32:30 - Collaboration Surfaces for Humans and AI
37:28 - Analysis of Jobs Market Data
48:25 - Open vs. Closed Source Models
53:51 - Autonomous Robotics and Atoms
1:00:59 - MicroGPT and Agentic Education
1:05:40 - End Thoughts

[Lydia Hallie ✨ @lydiahallie]

Claude Code on desktop lets you select DOM elements directly, much easier than describing which component you want updated!

Claude gets the tag, classes, key styles, surrounding HTML, and a cropped screenshot. React apps also get the source file, component name and props

[Zara Zhang @zarazhangrui]

Almost every AI power user I know is MORE stressed and busier after using AI, not less

What people thought AI would do: 10x productivity so that we can finish work earlier & relax more

What it’s actually doing: 10x productivity so that we end up with 20x more things to do cos of the sheer possibilities