Tuesday, October 21, 2025

new AI model: gpt-5-pro from OpenAI

GPT-5 pro

GPT-5 pro. Here's OpenAI's model documentation for their GPT-5 pro model, released to their API today at their DevDay event.

It has similar base characteristics to GPT-5: both share a September 30, 2024 knowledge cutoff and 400,000 context limit.

GPT-5 pro has maximum output tokens 272,000 max, an increase from 128,000 for GPT-5.

As our most advanced reasoning model, GPT-5 pro defaults to (and only supports) reasoning.effort: high

It's only available via OpenAI's Responses API.


generated this for price $1.1 
expensive!



 

distributed data SQLite sync: CRDTs (Conflict-Free Replicated Data Types).

Why Local-First Apps Haven’t Become Popular?

CRDTs (Conflict-Free Replicated Data Types).
guarantee two important properties:

Commutativity: Order of application doesn’t matter

Idempotence: Applying the same change twice has no effect

This means you can apply messages in any order, even multiple times, and every device will still converge to the same state.

SQLite Is Perfect for This
If you need a production-ready, cross-platform, offline-first engine for SQLite,
check out open-source SQLite-Sync extension.


GitHub - sqliteai/sqlite-sync: SQLiteSync is a local-first SQLite extension using CRDTs for seamless, conflict-free data sync and real-time collaboration across devices.

SQLite Sync is a multi-platform extension that brings a true local-first experience to your applications with minimal effort. It extends standard SQLite tables with built-in support for offline work and automatic synchronization, allowing multiple devices to operate independently—even without a network connection—and seamlessly stay in sync. With SQLite Sync, developers can easily build distributed, collaborative applications while continuing to rely on the simplicity, reliability, and performance of SQLite.

In simple terms, CRDTs make it possible for multiple users to edit shared data at the same time, from anywhere, and everything just works.


Monday, October 20, 2025

uv: fast Python tooling in Rust

 Reinventing Python tooling with Rust featuring Charlie Marsh from Astral (Changelog Interviews #660)

Charlie Marsh built Ruff (an extremely fast Python linter written in Rust) and uv (an extremely fast Python package manager written in Rust) because he believes great tools can have an outsized impact. He believes it so much, in fact, that he started an entire company that builds next-gen Python tooling


Sunday, October 19, 2025

AI Architecture: Round Passive Solar House

my "one-shot" "vibe designed" round house with Google Gemini Nano-Banana AI model

could win on some architecture design contests :)

"round passive solar house
with porches and decks on whole south side
and lighthouse-style sunroom on top of the roof;
use natural materials lodge-style;
solar panels on the roof"


"garages on the first floor"


round swimming pool covered with glass dome in front; with flowers all around


three-stores + lighthouse sunroom version


with round swimming pool covered with dome

back / north side of the house; natural materials; small windows with real wood shutters


and now some "real" round solar passive house

Villa Circuitus: Sweden's first round Passive House boasts an innovative solar-powered balcony


Clad in sustainable Kebony and orange panels, Villa Circuitus exudes a warm ambience that belies its smart design. Its round shape minimizes the surfaces exposed to the outside air, which, combined with high-performance custom windows that retain twice as much heat as standards windows, helps to minimize heat loss. As a result, less energy is required to keep the home warm in winter. Interior air temperatures are also kept stable with a heat exchanger that circulates air in a two-phase process – first through a ground-based water heater and then a post-heater.


Here's What You Need To Know About Passive Houses – Sunset Magazine


Round House Magic in the Mountains! He built a Wooden Yurt from a Kit - YouTube

Maple Twenty Five | 25-Foot Yurt Kit — Smiling Woods Yurts



Round Small House Design w/plans! Full Tour! - YouTube


QJS: JavaScript in Go with QuickJS & WASM

GitHub - fastschema/qjs: QJS is a CGO-Free, modern, secure JavaScript runtime for Go applications, built on the powerful QuickJS engine and Wazero WebAssembly runtime

QJS is a CGO-Free, modern, secure JavaScript runtime for Go applications, built on the powerful QuickJS engine and Wazero WebAssembly runtime.

QJS allows you to run JavaScript code safely and efficiently, with full support for ES2023 features, async/await, and Go-JS interoperability.




Thursday, October 16, 2025

AI HW: Apple M5 chip

Apple unleashes M5, the next big leap in AI performance for Apple silicon - Apple

M5 delivers over 4x the peak GPU compute performance for AI compared to M4, featuring a next-generation GPU with a Neural Accelerator in each core, a more powerful CPU, a faster Neural Engine, and higher unified memory bandwidth





AI Helping African Farmers

Agriculture & Food Systems Agentic AI News - Week Ending 2025-09-30 (Detailed)

"In Malawi, Africa, thousands of small farmers are using an AI chatbot created by Opportunity International. This AI agent gives farming advice through simple text messages. After Cyclone Freddy destroyed Alex Maere's corn farm in 2023, the AI agent suggested he grow potatoes.

Maere followed the AI's advice exactly and made over $800 selling potatoes. This money helped him pay for his children's school fees. The AI chatbot speaks local languages and gives advice that works in African growing conditions."



Wednesday, October 15, 2025

Top Programming Languages 2025

hew "real" and for whom is this statistics?

that knowing Python is 10x more important than HTML? simply no!

AI craze...

The Top Programming Languages 2025 - IEEE Spectrum

How to Lie with Statistics - Wikipedia



Tuesday, October 14, 2025

Nvidia $4K Mini PC

Elon Musk Gets Just-Launched NVIDIA DGX Spark: Petaflop AI Supercomputer Lands at SpaceX | NVIDIA Blog

“Imagine delivering the smallest supercomputer next to the biggest rocket,” 

 Nvidia to Start Selling $3,999 DGX Spark Mini PC This Week | PCMag

128GB RAM, 1000 TOPS

1 petaFLOP of AI performance at FP4 precision with the NVIDIA Grace Blackwell architecture.


A Grace Blackwell AI supercomputer on your desk | NVIDIA DGX Spark


SpaceX completes 11th test flight of its Starship rocket



AI vs Jobs? AuI "Augmented Intelligence" solution

(we) techno optimists still have hope... 

but it is "complicated..."

"artificial" machine based intelligence, whether called "General" (AGI) or "Super" (ASI)
can either "augment" (expand, support) human intelligence, or "replace".
The first is a "good" solution, the second one is clearly "bad".

Now, since history can be a guide, and while it may not "repeat" it may "rhyme", 
when mechanical machines are invented
they "augmented" human physical labor and unlocked "industrial revolution"

So we don't try to compete in speed with a train, car, airplane, we just "leverage" those tools.
And that works very well. "Augmented" mechanical power works.

For that reason I think we should start calling "AI = Augmented Intelligence" AuI
and treat it as such.

And the solution making this new AI power available for most people
should be some amount of free AuI tools. 
Like "free" roads... 

AGI Will Freeze the Economy - YouTube 
by Matthew Berman - YouTube

What Would Happen If Everyone Got $1,000 a Month? (UBI Theory Explained) - YouTube 
by Andrei Jikh - YouTube

Artificial Intelligence vs. Augmented Intelligence - YouTube (IBM)


Ex-OpenAI Scientist WARNS: You Have No Idea What's Coming #ArtificialIntelligence #AI - YouTube

Monday, October 13, 2025

AI: Agent Payments Protocol (AP2)

Announcing Agent Payments Protocol (AP2) | Google Cloud Blog

Google announced the Agent Payments Protocol (AP2), an open protocol developed with leading payments and technology companies to securely initiate and transact agent-led payments across platforms. The protocol can be used as an extension of the Agent2Agent (A2A) protocol and Model Context Protocol (MCP). In concert with industry rules and standards, it establishes a payment-agnostic framework for users, merchants, and payments providers to transact with confidence across all types of payment methods.






Tesla Robotaxi

 Tesla Robotaxi Reports 3 Crashes In Austin On One Day, Hides Details

Tesla’s crash report for self-driving vehicles to the federal government lists 3 crashes involving the Tesla robotaxi, all dated July 1, just ten days after launching their pilot service in Austin, TX. Tesla redacted much of the relevant data about the crashes, claiming they reveal important proprietary information. By this date, Tesla’s approximately dozen cars had done no more than 7,000 operational miles.



Sunday, October 12, 2025

Architecture: new Frank Lloyd Wright designed house

 See Frank Lloyd Wright’s Final Design, Revealed for the Very First Time | Architectural Digest

Wright’s design for a Usonian house he called Project #5909 had been on his drawing board when he died in 1959... brought to life (a new house) 65 years later, in 2025.


RiverRock House in Willoughby Hills, Ohio



AI: Gemini 2.5 Computer Use model from Google

 Introducing the Gemini 2.5 Computer Use model

Google is releasing the Gemini 2.5 Computer Use model via the Gemini API, enabling developers to build agents that can interact with user interfaces. This model outperforms others in web and mobile control benchmarks with lower latency. You can access it now on Google AI Studio and Vertex AI to start building and share feedback in the Developer Forum.

The model’s core capabilities are exposed through the new `computer_use` tool in the Gemini API and should be operated within a loop. Inputs to the tool are the user request, screenshot of the environment, and a history of recent actions. The input can also specify whether to exclude functions from the full list of supported UI actions or specify additional custom functions to include.



Saturday, October 11, 2025

Nvidia AI "intelligence" ecosystem investments

Nvidia company now has (by far) largest market valuation in the world.
And to sustain that growth, it is making strategic investments in "ecosystem"
including to "competitors" like Intel, customers like "OpenAI", partners like "ARM",
and most of all startups in all parts of "AI stack", like xAI, Cursor, CoreWeave...

Companies ranked by Market Cap - CompaniesMarketCap.com


Jensen Huang says he regrets not investing more in xAI: 'Almost everything Elon's part of, you want to be part of'

"My favorite enterprise AI service is Cursor," Huang said. "Every one of our engineers, 100%, is now assisted by AI coders, and our productivity has gone up incredibly."

"For the last several years, they were generating tokens at a loss," he said. "But in the last several months, it's become clear the technology is now reasoning, doing research, using tools — it's actually useful. The tokens are profitable now."

energy: Passive House City

the best way to get energy is to save energy, i.e. for heating/colling

"passive house" means house that does not need "active" heating/cooling

could a whole city district be build in such way, saving 90+% of energy?

yes!

A Passive House City? You better believe it! - YouTube

What exactly is a Passive House? | The Short of it! - YouTube

Home - Bahnstadt Heidelberg


GO INSIDE This 100% Passive House District in Germany - YouTube

Reimagine Buildings - YouTube  @lkll Passive House Accelerator : Catalyst for Zero Carbon Building

Passivhaus Institut



Passiv haus (foqushouse)


Friday, October 10, 2025

OpenAI SDKs: AgentKit Agent Builder (vs n8n)

OpenAI provides various SDKs for AI SW devs:

- standard AI API for request/response and chat, including support for "function calling/tools"
- Agents SDK with more advanced "AI Agents"
- AgentKit SDK integrated visual toolset for building Agents.

API Agents | OpenAI

OpenAI Agents SDK TypeScript | OpenAI Agents SDK

  • Enough features to be worth using, but few enough primitives to make it quick to learn.
  • Works great out of the box, but you can customize exactly what happens.
main features of the SDK:
  • Agent loop: Built-in agent loop that handles calling tools, sending results to the LLM, and looping until the LLM is done.
  • Handoffs: A powerful feature to coordinate and delegate between multiple agents.
  • Guardrails: Run input validations and checks in parallel to your agents, breaking early if the checks fail.

  • Sessions: Automatic conversation history management across agent runs, eliminating manual state handling.
  • Function tools: Turn any Python function into a tool, with automatic schema generation and Pydantic-powered validation.
  • Tracing: Built-in tracing that lets you visualize, debug and monitor your workflows, as well as use the OpenAI suite of evaluation, fine-tuning and distillation tools.

new tools for AI "workflows"

Intro to Agent Builder - YouTube

I Tested OpenAI's AgentKit Against n8n: What You Need to Know - YouTube

OpenAI Update: Building Agents using NEW Agent Builder! - YouTube


AgentKit, a complete set of tools for developers and enterprises to build, deploy, and optimize agents. 

  • Agent Builder: a visual canvas for creating and versioning multi-agent workflows
  • Connector Registry: a central place for admins to manage how data and tools connect across OpenAI products
  • ChatKit: a toolkit for embedding customizable chat-based agent experiences in your product



$12K hybrid car, 1300 miles range?

 The West Can't Compete: This $12,000 USD PHEV gets 1300 miles of range - YouTube

SAIC's Roewe M7 PHEV with 2,050 km range and ByteDance's AI model launched from 12,100 USD




Thursday, October 09, 2025

story: Google AI

WOW!
Even if someone thinks to be well informed about tech,
will for sure find new and valuable info in this amazing story.

Like AT&T and IBM before, Google is inventing the future. 
Most of what we call "AI" today directly or indirectly originated from Google.

Google: The AI Company: The Complete History and Strategy

"Google faces the greatest innovator's dilemma in history. They invented the Transformer — the breakthrough technology powering every modern AI system from ChatGPT to Claude (and, of course, Gemini). They employed nearly all the top AI talent: Ilya Sutskever, Geoff Hinton, Demis Hassabis, Dario Amodei — more or less everyone who leads modern AI worked at Google circa 2014. They built the best dedicated AI infrastructure (TPUs!) and deployed AI at massive scale years before anyone else. And yet... the launch of ChatGPT in November 2022 caught them completely flat-footed. How on earth did the greatest business in history wind up playing catch-up to a nonprofit-turned-startup?"



links
play with this here, free

TypeScript => DotNet?

While TypeScript compiler is being transpiled to GoLang, some apps are moving to C#...

Moving off of TypeScript. We Love You, TypeScript | by Chander Ramesh | Sep, 2025 | Motion Engineering Blog

"It’s no accident that Java and .NET rule the enterprise world. Runtime types are a given. Relatively fast compile times are mandatory for codebases that are millions of lines long. They both have an ORM that has matured over decades, which means for basic CRUD operations developers simply don’t need to think at all: there is only one real choice...."



Wednesday, October 08, 2025

AI Vibe coding with Replit

Very informative and some controversial views in this podcast: 
more demand for coding, more jobs and higher salaries, startups 

Replit CEO on The Career of Coding, AGI, and Vibe Coding Wars w/ Amjad Masad, Dave B & Salim - YouTube


Replit – Build apps and sites with AI


By embracing "Agentic AI" Replit valuation increased 3x in a year...

Replit (REPI.PVT) Valuation, History & News - Yahoo Finance


Replit's estimated valuation is around $3 billion, following a $250 million funding round in September 2025, which nearly tripled its value from the previous year.


WebAssembly/WASM 3.0

a big update, a big deal! very useful

Wasm 3.0 Completed - WebAssembly

WebAssembly Specifications


64-bit address space.

Multiple memories.

Garbage collection

Typed references

Tail calls.

Exception handling.

Relaxed vector instructions

Deterministic profile.

Custom annotation syntax

JS string builtins.



Tuesday, October 07, 2025

EV: Tesla Model Y and Model 3 Standard

2026 Tesla Model Y Standard: This Is It

Tesla has finally rolled out its long-promised affordable Model Y. The new base trim, called the Model Y Standard, is a stripped-down version of the company’s best-selling electric vehicle. It’s cheaper and simpler, but it still packs Tesla’s trademark tech, software and driving range found on pricier models.

 The Tesla Model Y and Model 3 Standard Are Cheaper—but Still Not Cheap | WIRED

Tesla announced a new Model Y and Model 3 Standard, versions of its popular compact SUV and sedan stripped of a few higher-end touches and features to bring the price down to $39,990 and $36,990, respectively. They’re both about $5,000 cheaper than the Premium variants


The Model Y has a new front and back profile, without a headlight- and taillight-spanning light bar. It comes with smaller, 18-inch wheels standard, and in only three colors. There’s less vegan leather and more textile on these seats, a reengineered roof that’s only glass on the outside, and no touchscreen in the rear passenger seats. The powertrain is less powerful—it adds 1.4 seconds to the Premium’s 0-to-60 time for the Standard, which comes in at 6.8 seconds—and the battery gives an estimated 321 miles of range per charge, compared to 363 for the higher-end model.



Chrome’s new AI features

 Go behind the browser with Chrome’s new AI features

 Gemini in Chrome 1 to Mac and Windows desktop users in the U.S. with their language set to English, so you can ask Gemini to clarify complex information on any webpage (or webpages) you're reading.



Monday, October 06, 2025

AI: AMD & OpenAI deal

 AMD signs AI chip-supply deal with OpenAI, shares surge over 34% | Reuters

  • AMD to supply OpenAI hundreds of thousands of AI chips
  • Deal includes warrant for OpenAI to buy up to 10% AMD stake at 1 cent/shr
  • AMD expects more than $100 billion revenue from OpenAI, others as a result of deal
  • AMD shares on track to record highest percentage gain since April 2016
OpenAI, which is valued at $500 billion, generated around $4.3 billion in revenue in the first half of 2025 and burned through $2.5 billion in cash, according to media reports.



Google GenAI textbooks

 Learn Your Way: Reimagining textbooks with generative AI

New research into GenAI in education demonstrates a novel approach to reimagining textbooks that led to improved learning outcomes in a recent study. The research comes to life in our interactive experience, Learn Your Way, now available on Google Labs.



Sunday, October 05, 2025

Sky AI delivery robots: Zipline

apparently already in USA, and much more in Africa

What is Amazon doing?

Zipline Drone Delivery & Logistics


Zipline - YouTube

Keller Rinaudo Cliffton (@Keller) / X

Science Corner Special! David Friedberg, Cleo Abram, Alex Filippenko, and Keller Rinaudo Cliffton - YouTube


AI MCP protocols: SSE vs Streamable HTTP

Note: SSE Transport is deprecated as of MCP specification version 2025-03-26, replaced by Streamable HTTP.

Why It's Being Replaced

SSE transport has three major limitations:

  1. No support for resumable streams
  2. Requires server to maintain long-lived, highly available connections
  3. Only allows server messages to be delivered via SSE (unidirectional)

Servers typically expose multiple endpoints:

  • /sse + /messages (legacy SSE)
  • /mcp (new Streamable HTTP)


MCP (Modal Context Protocol), one of the most popular and widely used AI protocols today, replaced the HTTP+SSE transport mechanism with Streamable HTTP starting from protocol version 2025-03-26. That has marked a significant change in the protocol’s architecture.

Specifically, MCP uses JSON-RPC 2.0 as its wire format between clients and servers. For transmission of JSON-RPC messages, it relies on standard transport mechanisms like HTTP+SSE or Streamable HTTP (among stdio — for communication over standard in and standard out on local servers).


Those specialized transport layers are necessary because traditional HTTP’s request-response model is inefficient for real-time AI communication. That is because plain HTTP introduces high overhead and latency due to frequent connection setups. In contrast, MCP requires continuous, low-latency data streams—something HTTP+SSE and Streamable HTTP are designed to handle.





GitHub - sidharthrajaram/mcp-sse: A working pattern for SSE-based MCP clients and servers

Course: The Complete Model Context Protocol (MCP) Masterclass | Udemy Business

Saturday, October 04, 2025

Nvidia, OpenAI: Perpetuum mobile money machine?!

As useful as it may be AI is not free.
In fact it is very expensive.
Both in money, and in electrical energy.
And there is not enough energy, not even close. Will get expensive.

Thanks to NVIDIA HW and AI models, in last a few years,
stock market is "betting" heavily on AI related stocks.
In fact this is what keep stock market afloat, preventing recession.

Now, for this "machine" to keep sucking more money, it needs to keep spending money.
Also, Nvidia now has "China problem", prevented to sell in volume to China.
Solution:
NVidia invests huge money in OpenAI, and OpenAI then spends that money buying from Nvidia.
Magic!
And to deploy all this HW, someone needs to build "data centers".
Since this customer is "secured", Oracle can get big loans to build and operate them. 
Magic again!
And since this is "good for economy, and good for country", politicians and bankers are on board.
Third time the magic!

But will all this "power" be able to deliver so much business value to justify investment?
Or this may end up being another "dot-com bust" gigantic Ponzi scheme

Nvidia’s $100B Deal with OpenAI: The Biggest AI Deal in History - YouTube

AI: A Perpetuum Mobile, or perpetual motion machine, is a hypothetical device that could operate indefinitely without any external energy input, typically by performing work continuously. However, such machines are impossible because they inevitably violate the fundamental Laws of Thermodynamics, which state that energy cannot be created out of nothing (First Law) and that no process is 100% efficient, as some energy is always lost as heat (Second Law). Throughout history, many designs have been proposed and built to demonstrate perpetual motion, but all have failed because they cannot overcome the inevitable losses due to friction and other factors.


  • Strategic partnership enables OpenAI to build and deploy at least 10 gigawatts of AI datacenters with NVIDIA systems representing millions of GPUs for OpenAI’s next-generation AI infrastructure. 
  • To support the partnership, NVIDIA intends to invest up to $100 billion in OpenAI progressively as each gigawatt is deployed.
  • The first gigawatt of NVIDIA systems will be deployed in the second half of 2026 on NVIDIA’s Vera Rubin platform.



Ai Agents vs Vibe coding

As to be expected, enthusiasm for "vibe" code gen based on unstructured spoken language didn't last.

Even with mostly automatic tools, one needs to "think". And that is hard for most people.

So now new "spin" is to delegate thinking/planning to AI tools, too.

The problem is not "artificial intelligence", it is "natural stupidity".

And the only cure for this is hard work, and thinking.

And no AI tool can help with that. But it can replace.

Vibe coding is dead: Agentic swarm coding is the new enterprise moat



Friday, October 03, 2025

AI: MCP, A2A, ACP Protocols

AI Agent Protocols Explained: MCP, A2A, ACP and More - YouTube

Message from AnswerRocket

FeatureMCPA2AACP
Primary PurposeConnects AI to tools and data sourcesConnects AI agents to other AI agentsEnables local agent coordination
Initiated ByAnthropicGoogle and partnersIBM/BeeAI
Best ForTool discovery and data accessMulti-agent collaboration across platformsEdge computing and local environments
Design FocusSimplifying context accessAgent interoperabilityLow-latency, offline operation
SecurityBasic; evolved from hobbyist projectEnterprise-grade by designLocal sovereignty, no cloud dependency
Communication StyleTool invocationTwo-way agent communicationEvent-driven messaging
Network RequirementsInternet/cloud-basedInternet/cloud-basedCan operate offline
Ideal EnvironmentEnterprise cloud applicationsCross-platform agent ecosystemsEdge devices, robotics, IoT



An Unbiased Comparison of MCP, ACP, and A2A Protocols | by Sandi Besen | Medium

MCP, ACP, A2A, Oh my! — WorkOS Guides

MCP vs A2A vs ACP: AI Protocols Explained | Bluebash


GitHub - Tsadoq/a2a-mcp-tutorial: A tutorial on how to use Model Context Protocol by Anthropic and Agent2Agent Protocol by Google

GitHub - GongRzhe/A2A-MCP-Server: A mcp server that bridges the Model Context Protocol (MCP) with the Agent-to-Agent (A2A) protocol, enabling MCP-compatible AI assistants (like Claude) to seamlessly interact with A2A agents.




Thursday, October 02, 2025

ICF vs SIPs vs Framing vs Factory house construction

Most of products are improving technology, while housing, is USA at least, is not changing much.
Hopefully that may change soon..
 
good, informative video of "traditional" methods
ICF vs SIPs vs Framing - Pros and Cons - YouTube



"Factory-build" houses. Like Tesla cars.
Would help if they hired some real good designer / architect. 
Most people don't like to live in a small "container" box.