Shashwat Jain retweeted
Introducing Reality-Aware Rendering in @Simulon, a completely new way to create with the real world.
Place neural assets with your phone and act out the scene. When you stop recording, Simulon renders the take, reading everything in it (what you do, what you say and the place around you), and your assets respond to all of it.
The final is regenerated from your recording: a new performance of the scene, not a copy of the footage.
Available now on iPhone. Coming to Android soon.
Gemini is redirecting 3.7 flash requests to 3.8 flash silently?? @GeminiApp
We are facing very high latency in our systems this is really absurd
4
210
Shashwat Jain retweeted
We're giving away a CONTROL Resonant custom wrapped NVIDIA GeForce RTX 5080 to celebrate its official release with #RTXON.
To enter:
🟢 Share this post
🟢 Comment #RTXON
T&Cs: nvda.ws/4z24Gcq
Single-most important thing that I have found important after working in startups for 5 years is the decision of how to choose a startup and it is actually really simple. Two major things:
1. You should be high agency and low ego like really low
2. See the founders and their decision making, no one should remain in a startup once they know that decision making is poor.
The call to leave the startup is actually easy but the difficult part is to let go of the investment of energy you would had done in case you are high agency which u might be if you are still reading this!!
This is insanneee!!! Computer use will be 10x better post this
Breaking: Browser Use + Jev = Ultrafast ⚡
Findings flights took 7s and cost only $0.0039 🤯
> new action space every step
> DOM state space
> small LLM fallback to type
(this video is at 1x speed btw)
Built a tiny open source browser agent. try it below ↓
1
59
This one is a big update, Achieving Pareto efficiency, is this GLM again? What are your thoughts nerds
#connect
Union Alpha (stealth model) is free for the next week
- no data training
- built for agentic coding
- supports images
let's see what you can do
58
Shashwat Jain retweeted
Say hello (literally) to Gemini 3.8 Live and 3.8 Live Extended Thinking, our new SOTA live audio models, available with frontier price + performance.
3.8 Live supports 97 languages (can seamlessly switch), async tool calls, and more!
Andrej Karpathy spent 8 years at OpenAI and Tesla
Last week, he condensed everything he knows into one free 2-hour lecture
Agents → Loops → Harness → Self-Improving Systems
People pay $14K for bootcamps that teach less than this, this lecture beats most paid AI engineering courses
Don't let this disappear from your feed, save it, watch it
2
156
Hello connections,
Wanted to understand your take on design what methodology do you apply for design
what is your perspective on ai-design, I want to build something and it seems design is the biggest bottleneck so any suggestions?
Also, happy to #connect
2
4
80
Any tech builders here?
I’m from India 🇮🇳
Into AI agents, software, DevOps, SaaS & open source.
Tell me your country + what you’re building.
Let’s #connect.
5
6
120
Hey all 👋
Looking to #connect with people that are building cool stuff and are interested in:
- SaaS
- Indie Hacking
- AI tools
- Automation
- Building in public
- following ai domain in detail
- useful tips and tricks
-using claude code/codex/muse
- using open source ai models
Let’s grow together 🤝
28
21
631
I’ve been building AI agents for ~2 years. Here are 7 things I believed about agents that I no longer believe.
8
2
74
I’m starting to write more about what actually works when building agentic systems — not agent demos.
Follow if you’re building in this space too.
#connect
1
1
27
The moat isn’t your prompt
It’s the surrounding system: data, tools, permissions, evaluations and feedback loops.
1
11
Observability becomes more important as models become smarter
Because smarter agents can fail through longer, more convincing trajectories.
1
7
Human-in-the-loop isn’t a temporary workaround
For many valuable workflows, approval boundaries are part of the product.
1
8
Most agent failures aren’t model failures
They’re state, permissions, bad tool schemas, retries and unclear termination conditions.
1
11
Memory isn’t primarily a vector DB problem
The difficult part is deciding what deserves to become memory.
1
11
Multi-agent is massively overused
A good single-agent loop often beats 5 agents talking to each other.
1
15












