📋 Log for 2026-03-19

😄 Joke of the Day

I used to hate facial hair, but then it grew on me.

Category: dad

YouTube Summaries

[Delegate to AI but never lose authority! #ai #futureofwork](https://www.youtube.com/shorts/e2_n7chCOhw)

Channel: NateBJones

Summary:

• Here's a summary of the video based on the provided text:

Key Takeaways

• The perceived need for engineers to simply code faster due to AI is an oversimplification; the reality of technical skills is more complex.
• The core shift in leverage is from writing code to orchestrating AI systems ("probabilistic systems").
• A fundamental "phase transition" from authorship to orchestration is occurring.
• This transition invalidates the traditional assumption that increased effort directly correlates with increased output.
• A new "four-level skill tree" is emerging, applicable to all roles, focusing on intent conditioning, context setting, and leveraging feedback loops, evaluations, and governance for compounding results.
• Separating the AI's "generation" capability from human "decisioning" is crucial for maintaining accountability and control.
• Authority in the new landscape stems from understanding and managing this inverted abstraction stack, where old technical boundaries are dissolving.
• Organizations that proactively develop skill trees around separating AI generation from human decision-making will achieve significant speedups (10x), while those maintaining old hierarchies will be outpaced.

Main Arguments

• The evolution of AI necessitates a redefinition of technical skills, moving beyond mere coding proficiency to encompass the strategic orchestration of AI tools.
• The traditional value proposition of effort-to-output is breaking down as AI takes on generative tasks.
• A structured approach to developing new competencies (the "skill tree") is essential for individuals and organizations to adapt.
• Accountability for AI-generated outputs requires a clear distinction between AI's generative capacity and human oversight/decision-making.
• Organizational structure and skill development strategies will determine success in the age of AI, with adaptability and a focus on new paradigms being key differentiators.

Notable Quotes

• "Andrej Karpathy says he feels behind."
• "The leverage has shifted from writing code to orchestrating probabilistic systems."
• "The phase transition from authorship to orchestration."
• "The old assumption that effort maps to output."
• "Conditioning intent and context."
• "Compounding through evals, feedback loops, and governance."
• "Separating generation from decisioning."
• "The one accountable for what the LLM produces."
• "Authority comes from... the abstraction stack got inverted and old technical boundaries no longer make sense."
• Organizations that build deliberate skill trees around separating generation from decisioning will realize "10X speedups."

Important Nuances

• The shift is framed as a "leveling crisis" where traditional technical hierarchies may become obsolete.
• The "skill tree" concept implies a structured, deliberate development path rather than organic skill acquisition.
• The idea of an "inverted abstraction stack" suggests a fundamental change in how complex systems are understood and managed, with high-level decisions influencing low-level AI behavior more directly.
• The emphasis on the year 2026 highlights an imminent critical period for organizations to adapt their strategies.

Published: 2026-03-19T21:00:52+00:00

[Perplexity Computer Is Incredible. It Won't Matter. Here's Why.](https://www.youtube.com/watch?v=3FbqaD1MCUA)

Channel: NateBJones

Summary:

• Here's a summary of the video based on the provided transcript and description:

Key Takeaways

• Perplexity's Dilemma: Despite Perplexity shipping what the author considers the best "agentic product of the month" (Perplexity Computer), its long-term viability is questioned because its core reasoning engine relies on models from direct competitors who are also developing similar products.
• Middleware Squeeze: Companies operating in the "middleware" layer of the AI stack face significant pressure from both below (model providers like OpenAI, Google) and above (context platforms), creating a challenging market environment.
• Strategic Positioning is Crucial: The argument is that excellent execution on a product can be insufficient if the company is positioned on the "wrong layer" of the technology stack, especially as larger players ("hyperscalers") become more dominant and hungry for control over tokens and data.
• Four Surviving Positions: The video outlines four structural positions that are predicted to survive and thrive in the evolving AI landscape:
• 1. Platforming Context: Mastering and controlling the specific context that agents require.
• 2. Infrastructure for Agents: Serving as the foundational infrastructure that agents call upon.
• 3. Owning the Customer Workflow: Deeply embedding into and controlling the end-to-end workflows of users.
• 4. Owning the Trust and Verification Layer: Being the layer responsible for trust and verification of information.
• Perplexity's True Strategic Asset: The video posits that Perplexity's "search API" is its real strategic advantage and "out," rather than its "Computer" product, suggesting this API represents a more defensible infrastructure play.
• Market Shift in February 2026: The date February 2026 is highlighted as a point that "hardened the demand signal" and clarified which players operate across multiple layers of the AI stack.

Main Arguments

• Execution vs. Layer: The primary argument is that good product execution alone cannot save a company if it's positioned on an unsustainable layer of the technology stack, particularly when its foundational dependencies (AI models) are controlled by aggressive competitors.
• Competitive Threat from Model Providers: The AI market is characterized by model providers (hyperscalers) who are not only providing the core intelligence but are also building the very applications that compete with their customers. This creates a fundamental conflict of interest.
• Consolidation and Pressure: The AI market is likely to consolidate, with pressure mounting on middleware companies as large players aim to capture more value, especially "tokens" (referring to the computational units or data processed by models).
• Defensible Positions in AI: The video suggests that to survive, companies must occupy one of the four identified strategic positions that offer inherent defensibility and alignment with future market trends.

Notable Quotes/Strong Statements

• "good execution on the wrong layer of the stack will not save you"
• "every model provider you depend on is simultaneously building the exact product you compete with."
• "Hyperscalers are coming for the tokens."
• Perplexity's "search API, not Computer, is their real strategic out."

Important Nuances

• Critique of Orchestration Moats: The video implicitly critiques the common notion that sophisticated "multi-model orchestration" is a sufficient competitive moat, arguing that it's less relevant when the underlying models themselves are commoditized or controlled by competitors.
• Timeline Significance: The mention of "February 2026" suggests a specific inflection point in the market where strategic advantages and vulnerabilities became more apparent.
• Strategic Depth: The analysis goes beyond product features to focus on deeper architectural and market positioning strategies for long-term survival in the AI space.
• "Agentic Product" Definition: The discussion revolves around "agentic products" which implies AI systems that can perform tasks autonomously or semi-autonomously, highlighting Perplexity Computer's capabilities in this domain.

Published: 2026-03-19T14:50:29+00:00

[Analysing The World's First Solid-State Battery](https://www.youtube.com/watch?v=WNX47A8BysE)

Channel: ZirothTech

Summary:

• Here's a summary of the video "Analysing The World's First Solid-State Battery" based on the provided description:

Key Takeaways

• A solid-state battery claimed by "Donut Lab" is causing excitement in the clean energy sector, with potential to revolutionize electric vehicles and energy storage.
• The presenter harbors significant doubts, suspecting that the full and accurate picture of the battery's capabilities is not being presented.

Main Arguments

• The video will involve a technical analysis, specifically by "Calculating the Curves" and "Comparing the Curves" of the battery's performance data.
• The presenter intends to confirm that the battery's composition does not involve sodium.
• A crucial, unidentified component referred to as "The Missing Element" is highlighted as central to the analysis.
• The core of the video appears to be an investigation into the veracity of the claims, questioning who might be misleading others ("Who is Fooling Who?").

Notable Quotes

• "I have a strong suspision we're not getting the full picture."
• "...promising to turn the world of electric vehicles and energy as a whole upside-down."
• "It's Not Sodium."
• "The Missing Element"
• "Who is Fooling Who?"
• "My Final Warning"

Important Nuances

• The analysis is to be performed without relying on visual content, focusing strictly on data and technical details.
• The presenter's critical stance suggests a potential exposé or a debunking of sensationalized claims surrounding the battery technology.
• The investigation will focus on specific technical aspects like performance curves and material composition.
• The video concludes with a strong cautionary note from the presenter.

Published: 2026-03-19T11:30:28+00:00

[AI agents work while you rest—if you do this #ai #futureofwork #agenticai](https://www.youtube.com/shorts/S8DKnN61WTA)

Channel: NateBJones

Summary:

• Here's a summary of the video content:

Key Takeaways

• AI agents claiming completion prematurely is a significant issue that smarter models alone may not fully resolve.
• The reliability of AI agents depends heavily on their design and evaluation loops, not just raw model capability.
• Defining clear criteria for "done" is crucial for enabling AI agents to iterate autonomously towards correct outcomes.
• The future of AI agent utilization, particularly for non-technical knowledge work, will focus on agentic harness design and workflow-shaped evaluations.

Main Arguments

• The expectation that simply using more advanced LLMs will fix AI agent premature completion is an oversimplification. The reality is more intricate, involving how agents are built and tested.
• Implementing a "simple eval loop" (evaluation loop) can effectively compel LLMs to refine their outputs and converge on correctness.
• The primary bottleneck in AI agent development is shifting from the model's inherent capabilities to the design of the "agentic harness," which structures and guides the agent's operation.

Notable Quotes/Phrases

• "Claude Code's biggest weakness is saying it's finished prematurely."
• "a simple eval loop forces LLMs to converge on correctness"
• "workflow-shaped evaluations mean for non-technical knowledge work in 2026"
• "the bottleneck is shifting from model capability to agentic harness design"
• "2026 belongs to people who can define what done looks like clearly enough that agents can iterate toward it autonomously."

Important Nuances

• The issue of premature completion highlights a systemic problem rather than just an LLM limitation.
• "Workflow-shaped evaluations" suggest a move towards assessing agents based on their ability to execute complex, multi-step processes, not just single-point outputs.
• The ability to clearly articulate task completion criteria will become a key skill for knowledge workers interacting with AI agents in the near future.

Published: 2026-03-19T03:00:28+00:00

Latest OpenRouter Models

Google: Gemma 4 26B A4B (google/gemma-4-26b-a4b-it)

Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. Despite 25.2B total parameters, only 3.8B activate per token during inference — delivering near-31B quality at a fraction of the compute cost. Supports multimodal input including text, images, and video (up to 60s at 1fps). Features a 256K token context window, native function calling, configurable thinking/reasoning mode, and structured output support. Released under Apache 2.0.

Published: 03/04/2026

https://openrouter.ai/google/gemma-4-26b-a4b-it

Google: Gemma 4 31B (google/gemma-4-31b-it)

Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output. Features a 256K token context window, configurable thinking/reasoning mode, native function calling, and multilingual support across 140+ languages. Strong on coding, reasoning, and document understanding tasks. Apache 2.0 license.

Published: 02/04/2026

https://openrouter.ai/google/gemma-4-31b-it

Qwen: Qwen3.6 Plus (free) (qwen/qwen3.6-plus)

Qwen 3.6 Plus builds on a hybrid architecture that combines efficient linear attention with sparse mixture-of-experts routing, enabling strong scalability and high-performance inference. Compared to the 3.5 series, it delivers major gains in agentic coding, front-end development, and overall reasoning, with a significantly improved “vibe coding” experience. The model excels at complex tasks such as 3D scenes, games, and repository-level problem solving, achieving a 78.8 score on SWE-bench Verified. It represents a substantial leap in both pure-text and multimodal capabilities, performing at the level of leading state-of-the-art models.

Published: 02/04/2026

https://openrouter.ai/qwen/qwen3.6-plus

Free Models Catalog

Xiaomi: MiMo-V2-Omni (xiaomi/mimo-v2-omni)

MiMo-V2-Omni is a frontier omni-modal model that natively processes image, video, and audio inputs within a unified architecture. It combines strong multimodal perception with agentic capability - visual grounding, multi-step planning, tool use, and code execution - making it well-suited for complex real-world tasks that span modalities. 256K context window.

Published: Wed, 18 Mar 2026 19:55:03 GMT

Xiaomi: MiMo-V2-Pro (xiaomi/mimo-v2-pro)

MiMo-V2-Pro is Xiaomi's flagship foundation model, featuring over 1T total parameters and a 1M context length, deeply optimized for agentic scenarios. It is highly adaptable to general agent frameworks like OpenClaw. It ranks among the global top tier in the standard PinchBench and ClawBench benchmarks, with perceived performance approaching that of Opus 4.6. MiMo-V2-Pro is designed to serve as the brain of agent systems, orchestrating complex workflows, driving production engineering tasks, and delivering results reliably.

Published: Wed, 18 Mar 2026 19:54:03 GMT

Robot Technology

🤖 How coffee manufacturers are automating palletizing (with payback under 1 Year)

Coffee production is growing worldwide. From roasted beans to capsules and pods, manufacturers are increasing output to meet rising demand. Capsule formats such as Dolce Gusto® and other single-serve systems continue to expand in both retail and food service. But while packaging technology has evolved quickly, many coffee plants still rely on manual palletizing at the end of the line . That’s becoming increasingly difficult to sustain. Operators are harder to hire. Labor costs are rising. And pa...

Source: blog.robotiq.com • Published: Thu, 19 Mar 2026 13:00:06 GMT

Read more

🤖 Overcoming Core Engineering Barriers in Humanoid Robotics Development

A technical examination of the sensing, motion control, power, and thermal challenges facing humanoid robotics engineers — with component-level design strategies for real-world deployment. What Attendees will Learn Why motion control remains the hardest unsolved problem — Explore the modelling complexity, real-time feedback requirements, and sensor fusion demands of maintaining stable bipedal locomotion across dynamic environments. How sensing architectures enable perception and safety — Underst...

Source: spectrum.ieee.org • Published: Thu, 19 Mar 2026 10:00:05 +0000

Read more

Good News

Gene from High Altitude Yak Protected and Repaired Myelin Sheath in Early MS Study

A special gene that helps animals like the yak survive at high altitude could enable new treatments for multiple sclerosis after positive findings from a mouse model. The genetic mutation that enables yaks to live in environments with much less oxygen, and may hold the key to repairing nerve damage in conditions such as multiple […] The post Gene from High Altitude Yak Protected and Repaired Myelin Sheath in Early MS Study appeared first on Good News Network .

Published: Wed, 18 Mar 2026 18:30:23 +0000

Read more

Elusive Nightjar Populations Doubled in 5 Years, a ‘Remarkable Comeback’ Conservationists Say

The population of one of England’s most-elusive birds is flourishing again thanks to conservation efforts in the south of the country. The nightjar, sometimes called the nighthawk, becomes active at twilight, and they’re famous for their chortling calls and fantastic camouflage. Their numbers declined some 51% between the 1970s and 2000s after substantial forest loss. The […] The post Elusive Nightjar Populations Doubled in 5 Years, a ‘Remarkable Comeback’ Conservat...

Published: Thu, 19 Mar 2026 13:00:00 +0000

Read more

Dialing ‘867-5309’ Will Now Connect You with a Professional Cancer Support Line Free of Charge

If you call 867-5309 these days, Jenny doesn’t answer, but a telemedicine professional will. After seeing the emotional toll cancer took on some of his friends, the man behind Tommy Tutone’s 1981 smash hit “867-5309/Jenny,” has teamed up with global nonprofit Cancer Support Community to turn the most famous phone number in music into a […] The post Dialing ‘867-5309’ Will Now Connect You with a Professional Cancer Support Line Free of Charge appeared fir...

Published: Thu, 19 Mar 2026 11:00:32 +0000

Read more

Good News in History, March 19

20 years ago today, the Second World Congress of Imams and Rabbis for Peace took place with 150 Jewish Rabbis and Muslim Imams aiming for Islam and Judaism to become instruments of peace. The goals of the first meeting in 2005 included 1) condemning violence perpetrated in the name of religion 2) the creation of […] The post Good News in History, March 19 appeared first on Good News Network .

Published: Thu, 19 Mar 2026 07:00:00 +0000

Read more

Amid energy market turmoil, the people taking power into their own hands

Communities are showing how clean power can be local, affordable and owned by the people who use it The post Amid energy market turmoil, the people taking power into their own hands appeared first on Positive News .

Published: Thu, 19 Mar 2026 07:00:12 +0000

Read more

← back to index