It quite surprised me that it's IBM doing this :) For anyone lost in the sea of terminologies, principles and needs guidance (like me), I recommend the channel bellow. IBM tutors go through different subjects in a clear and short way. https://youtube.com/@ibmtechnology?si=Mfq1iP1VVHlTf8b2

I’ve been running open-weight models locally for a while. GLM-5.2 on a 512GB Mac Studio was capable, but you always felt the ceiling: 2-bit quant, 5-9 tok/s, tool calls that got flaky in long agent sessions. Local was the thing you tolerated to keep data on your network. DeepSeek V4 Flash was the first real step forward — finally a local model with a genuine balance of performance and speed, one you could actually leave running under agent traffic without babysitting it. It proved the “fast MoE with real agentic chops” formula worked. 5.3-Flash on two DGX Sparks takes that further. 320B params but only 18B active, so it’s running at good speeds. 1M context. Thinking is always on but tunable (low/high/max), and at max it holds up in real agent work through Hermes: multi-step coding, sustained tool calls, no hand-holding. I’m still reaching for cloud models. But for the first time I can see a point coming where you won’t have to. The gap is closing faster than I expected — if you have the hardware, this is the one to try.

https://youtube.com/live/AAw5r0e-ozM?feature=share

Tech debt. After nearly a year of developing my niche, highly complex voice-and-AI-native software, I am declaring a RESET day. I am unsure whether to celebrate or cry. I am archiving my entire codebase and Supabase schema, choosing to rebuild rather than struggle with refactoring the heavy load of tech debt in the voice-first system I began last October following four months of full-time research. I am shifting from Supabase Cloud to Postgres with PITR and WAL-G, relocating self-hosted agents and communication to memory-optimized Azure Virtual Machines for the LiveKit real-time voice streaming cluster and agent engines. I am archiving the 1.5 million lines of code for a simpler structure, likely using Tailwind CSS with Next.js, React, and Tanstack Query. I am also moving Vercel hosting, Resend, and Twilio to a Canadian Azure VPS and ACS. I chose Azure because they supported my startup with a generous chance to accelerate without bankruptcy :) and without requiring VC funding for higher support tiers, unlike the other major players (I am looking at you, @googledevs and @awscloud) - if you change your mind, reach out! ;) **So tonight, it is a fresh start.** ——— Thanks to @Ray Fernando and community - that last Builder Room conversation got me thinking… and then, seeing how my agents have evolved sooo much in two months, it feels like the right time to start fresh, and ship fast!

I sat on a working app for a month because I was waiting for a perfect plan. SayRay is live captions on your phone. Apple's new models in iOS 27. On device. Real-time translation. No servers. I am shipping the prototype to the App Store for five dollars. This is day 1 of 180. Watch: https://youtu.be/h0M-N-JCYIo If you want to go through the same kind of ship with me: https://rfer.me/startmyai