ATP

@ATPinsights

Michael breaks down the most important tech news in Asia, and brings experts to share their insights.

Singapore

youtube.com/@asiatechpodca…

Replies Reposts Media

Pinned
ATP
@ATPinsights
Aug 31
ATP Insights #54 - On Today's Show: 1. Nvidia Wants to Run the World’s Robots 2. Tencent Unveils Its New Hy4 AI Model 3. Bryan Sng, Co-Founder and COO at @simpleai_sg 4. Amanda Yeo, Founder & Managing Partner at Brighten Sphere 5. @thurupathanv, Co-Founder / CEO at @aventude 6.
ATP
@ATPinsights
Aug 31
x.com/i/broadcasts/1…
ATP
@ATPinsights
6m
Tencent just shrank Hy4 Preview to 214GB today. Here's what you need to know. Tencent says its Hy4 preview model went from 1.5TB to 214GB, about seven times smaller, using Sherry, a quantization method that packs weights to 1.25 bits while keeping quality close to BF16. Hy4
ATP
@ATPinsights
52m
Whatfix co-founder Khadim Batti died suddenly today. Here's what you need to know. Batti, co-founder and CEO of Whatfix, passed away on Tuesday, Sep. 1, 2026, following a cardiac arrest. Avinash Raghava posted the news at 5:42 PM, saying Batti was central to AIBoomi and India’s
Avinash Raghava 🇮🇳
@avinashraghava
2h
I’m deeply shocked and saddened by the passing of Khadim Batti, Co-founder of Whatfix, earlier today. I'm truly struggling to find the right words while processing this news. Khadim has been such an important part of our AIBoomi community and of the larger SaaS and AI ecosystem
ATP
@ATPinsights
1h
Hugging Face just sparked an AI-agent oversight debate today. Here's what you need to know. A Sep. 1 post by Alex Prompter framed a new Hugging Face-linked position paper as a warning that AI agents can make people worse overseers. The argument is simple, the more an agent does,
Alex Prompter
@alex_prompter
6h
Hugging Face just argued AI agents make the humans using them worse at their jobs. Here's what their paper found in plain English: The more your agent does, the less you do. You stop doing the work and start approving it. Over months that changes how you work: approving faster
ATP
@ATPinsights
2h
Abliteration.ai just released a guardrail-free version of GLM-5.3 today. Here's what you need to know. Abliteration.ai launched "abliterated-model-large-v2," a stripped version of Z.ai's GLM-5.3 model with its safety layer removed. The company
Abliteration.ai
@abliteration_ai
21h
Today we're releasing abliterated-model-large-v2. Based on GLM-5.3, which is #3 on Terminal-Bench 4.0 (behind only Opus 5 and Fable), with 2× the cyber exploitation of 5.2. We abliterated and hosted it so it does the offensive cyber, red teaming, and agent testing work other