Author | Lightnews

Kuan Hoong @kuanhoong.bsky.social · 18d

AWS User Group AI/ML, Kuala Lumpur www.meetup.com/awsug-aiml-my

Kuan Hoong @kuanhoong.bsky.social · 19d

Annual contributions as AWS Community Builders in AI&ML at AWS Community Day Malaysia #aws #awscommunitybuilders #awscommunityday

microsoft.ai/news/introdu...

Kuan Hoong @kuanhoong.bsky.social · 19d

AI Agent Trends of 2025

Kuan Hoong @kuanhoong.bsky.social · 19d

Introducing MAI-Image-1, debuting in the top 10 on LMArena | Microsoft AI

Introducing MAI-Image-1, debuting in the top 10 on LMArena | Microsoft AI

Kuan Hoong @kuanhoong.bsky.social · 19d

With high-speed, high-quality output and planned integration into flagship tools, MAI-Image-1 could soon empower millions of users with creative AI capabilities built directly into their workflows shifting control, innovation, and differentiation in the competitive AI ecosystem.

Introducing MAI-Image-1, debuting in the top 10 on LMArena | Microsoft AI

Kuan Hoong @kuanhoong.bsky.social · 19d

Why this matters
This move marks Microsoft’s growing push to build its own generative media stack rather than relying fully on external partners. Cracking the LMArena top 10 on release signals competitiveness in the image generation space.

Introducing MAI-Image-1, debuting in the top 10 on LMArena | Microsoft AI

Kuan Hoong @kuanhoong.bsky.social · 19d

MAI-Image-1 is optimized for speed and visual fidelity, particularly in photorealistic rendering of landscapes, lighting, and reflections. Microsoft intends to integrate it into Copilot and Bing Image Creator soon, while continuing to refine safety and output variability.

Introducing MAI-Image-1, debuting in the top 10 on LMArena | Microsoft AI

Kuan Hoong @kuanhoong.bsky.social · 19d

Microsoft releases MAI-Image-1 to compete with Google's Nano Banana

Microsoft has unveiled MAI-Image-1, its first in-house text-to-image generation model. On debut, it made it into the top 10 on the LMArena leaderboard, tying with models like Seedream 3.

Now open for building: Introducing Gemini CLI extensions

Kuan Hoong @kuanhoong.bsky.social · 23d

My YouYube Channel Monthly Stats.
Subscribe to my channel youtube.com/@kuanhoong

Kuan Hoong @kuanhoong.bsky.social · 24d

Google AI Plus is here too for Malaysians RM24/month

support.google.com/googleone/an...

Kuan Hoong @kuanhoong.bsky.social · 24d

Circular economy or AI bubble ?

2

Kuan Hoong @kuanhoong.bsky.social · 25d

Over time, the extension ecosystem can become a force multiplier for both productivity and innovation across developer communities.

blog.google/technology/d...

Make Gemini CLI uniquely yours by connecting it to your everyday workflows and the tools you use most.

Now open for building: Introducing Gemini CLI extensions

Kuan Hoong @kuanhoong.bsky.social · 25d

In the race of agentic development tools (like Claude Code, Codex, Gemma CLI), this modularity gives Gemini a strategic edge. It accelerates adoption by letting teams embed AI into existing pipelines without reinventing their stack.

Make Gemini CLI uniquely yours by connecting it to your everyday workflows and the tools you use most.

Now open for building: Introducing Gemini CLI extensions

Kuan Hoong @kuanhoong.bsky.social · 25d

Why this matters

Extensions turn Gemini CLI from a generic tool into a deeply personalized agent, one that understands your environment, automates context-aware tasks, and interacts with your preferred services seamlessly.

Make Gemini CLI uniquely yours by connecting it to your everyday workflows and the tools you use most.

Now open for building: Introducing Gemini CLI extensions

Kuan Hoong @kuanhoong.bsky.social · 25d

Developers can package custom commands, context instructions, and tool connections into extensions, letting Gemini CLI operate within their tailored workflows.

Make Gemini CLI uniquely yours by connecting it to your everyday workflows and the tools you use most.

Now open for building: Introducing Gemini CLI extensions

Kuan Hoong @kuanhoong.bsky.social · 25d

Extensions are installable modules that include a “playbook” guiding the AI to use connected tools effectively from day one. These integrations span databases, design platforms, payment systems, cloud services, security tools, and more.

Make Gemini CLI uniquely yours by connecting it to your everyday workflows and the tools you use most.

Now open for building: Introducing Gemini CLI extensions

Kuan Hoong @kuanhoong.bsky.social · 25d

Gemini CLI extensions

Google has introduced Gemini CLI extensions, allowing users to customize their command-line AI agent by integrating it with tools and services they already use.

Make Gemini CLI uniquely yours by connecting it to your everyday workflows and the tools you use most.

OpenAI will let developers build apps that work inside ChatGPT

Kuan Hoong @kuanhoong.bsky.social · 27d

Build apps inside ChatGPT www.theverge.com/news/793039/...

Ask Zillow inside ChatGPT for help finding a home.

www.theverge.com

Kuan Hoong @kuanhoong.bsky.social · Oct 1

Until Oct 6, all Gemini usage through LOVEABLE is completely FREE! 🎉 lovable.dev/blog/free-lo...

Kuan Hoong @kuanhoong.bsky.social · Oct 1

Watch the launch video: www.youtube.com/watch?v=gzne...

iOS app: apps.apple.com/us/app/sora-... (Only available in US and Canada for now)

Introducing Sora 2

YouTube video by OpenAI

www.youtube.com

Kuan Hoong @kuanhoong.bsky.social · Oct 1

The launch of the companion app (iOS) with identity control, moderation, and consumption safeguards—also reflects how deployment decisions are becoming as critical as model performance in this evolving era of immersive AI experiences.

openai.com/index/sora-2/

Sora 2 is here

Our latest video generation model is more physically accurate, realistic, and more controllable than prior systems. It also features synchronized dialogue and sound effects. Create with it in the new ...

Our latest video generation model is more physically accurate, realistic, and more controllable than prior systems. It also features synchronized dialogue and sound effects. Create with it in the new ...

Kuan Hoong @kuanhoong.bsky.social · Oct 1

As video and audio generation move from novelty toward reliable tools, Sora 2 could reshape media creation, social storytelling, virtual environments, and human–machine interaction.

Sora 2 is here

Our latest video generation model is more physically accurate, realistic, and more controllable than prior systems. It also features synchronized dialogue and sound effects. Create with it in the new ...

Kuan Hoong @kuanhoong.bsky.social · Oct 1

where users can create, remix and insert themselves into scenes via “cameos,” with built-in controls to support wellbeing and content moderation.

Why this matters
Sora 2 brings us a step closer to seamless world simulation—AI that understands and replicates physics, continuity, and human presence.

Sora 2 is here