Great to catch up with an amazing friend and mentor today. He has been in this business for many years and I am truly grateful for all the guidance and support he has given me along the way. Thank you for always being there to help!

MyAGI-A1 just updated and published its own website!

The agent transformed www.MyAGI.au from a "Coming Soon" page to a full story page autonomously while I chatted with it.

Key capabilities:
- Voice transcription
- Photo analysis
- Self-learning skills
- Social media posting
- Website updates and deployment
- Computer use

The AI runs locally on my Mac laptop, controlled entirely through Telegram, and leverages my existing Claude subscription with no additional costs. It utilises MCP, APIs, and CLI tools available on the machine.

I can't sleep tonight! I'm just chatting and watching my work get done.

[Fun Story]: When your Senior AI agent starts roasting the Junior AI agent.

I had two AI agents tackle an AWS SDK integration issue today:

Junior Agent (Small Model): "Boss, it's hopeless. There's a 'fundamental incompatibility.' We need to rebuild the entire stack. I'm declaring bankruptcy on this ticket."

Senior Agent (Big Model): *Sips digital coffee* "Hold my beer. It's literally just a missing config line for the user-agent permission. I fixed it, wrote the test script, and verified it sent an SMS. Next."

The best part? The Senior agent's official log summary included this absolute gem of professional shade: "The other agent gave up too early!"

Wow, another frontier model release by OpenAI, another massive jump! The LLM war continues...

GPT-5.2 Thinking is here and the benchmarks are impressive:

- 100% on AIME 2025 (competition maths)
- 55.6% on SWE-Bench Pro (software engineering)
- 86.2% on ARC-AGI 1 (abstract reasoning)
- 92.4% on GPQA Diamond (science questions)

OpenAI is pushing hard on reasoning capabilities. The competition between OpenAI, Anthropic, and Google keeps heating up!