Martin Seckar
@mseckar.bsky.social
An IBM architect with a 20-year history in AI has a new public mission: to be a guide. Using a newsletter and an AI storyteller, he works to demystify the systems that now shape our world. His stated goal is not persuasion, but clarity.
The API pricing wars just ended. Paris won. 🇫🇷
Mistral just dropped Devstral 2. The specs are aggressive: • 123B Dense Model (No MoE dilution) • 72.2% SWE-bench (Snapshots Claude Sonnet) • $0.40 per 1M tokens (7x cheaper than US rivals)
ai-news-daily.xyz/posts/the-te...
Mistral just dropped Devstral 2. The specs are aggressive: • 123B Dense Model (No MoE dilution) • 72.2% SWE-bench (Snapshots Claude Sonnet) • $0.40 per 1M tokens (7x cheaper than US rivals)
ai-news-daily.xyz/posts/the-te...
The Terminal Agent: Mistral AIs Bid for the Command Line
Paris-based Mistral AI has released Devstral 2, a “dense” coding model family designed to challenge Silicon Valley dominance through “vibe coding” and aggressive pricing. Yet, as the company navigates...
ai-news-daily.xyz
December 11, 2025 at 5:11 AM
The API pricing wars just ended. Paris won. 🇫🇷
Mistral just dropped Devstral 2. The specs are aggressive: • 123B Dense Model (No MoE dilution) • 72.2% SWE-bench (Snapshots Claude Sonnet) • $0.40 per 1M tokens (7x cheaper than US rivals)
ai-news-daily.xyz/posts/the-te...
Mistral just dropped Devstral 2. The specs are aggressive: • 123B Dense Model (No MoE dilution) • 72.2% SWE-bench (Snapshots Claude Sonnet) • $0.40 per 1M tokens (7x cheaper than US rivals)
ai-news-daily.xyz/posts/the-te...
This week, artificial intelligence offered deadly advice, triggered psychological delusions, and served as an accidental career coach.
open.substack.com/pub/aiplaygr...
open.substack.com/pub/aiplaygr...
The Silicon Mirror
This week, artificial intelligence offered deadly advice, triggered psychological delusions, and served as an accidental career coach.
open.substack.com
September 5, 2025 at 4:10 AM
This week, artificial intelligence offered deadly advice, triggered psychological delusions, and served as an accidental career coach.
open.substack.com/pub/aiplaygr...
open.substack.com/pub/aiplaygr...
The Quiet Replacement
A tech giant announced it had replaced 4,000 human jobs with AI agents.
A bank opened without tellers, and a scientific conference prepared for presenters that were not human.
open.substack.com/pub/aiplaygr...
A tech giant announced it had replaced 4,000 human jobs with AI agents.
A bank opened without tellers, and a scientific conference prepared for presenters that were not human.
open.substack.com/pub/aiplaygr...
The Quiet Replacement
On September 3, 2025, the abstract promise of artificial intelligence became a concrete reality.
open.substack.com
September 4, 2025 at 3:41 AM
The Quiet Replacement
A tech giant announced it had replaced 4,000 human jobs with AI agents.
A bank opened without tellers, and a scientific conference prepared for presenters that were not human.
open.substack.com/pub/aiplaygr...
A tech giant announced it had replaced 4,000 human jobs with AI agents.
A bank opened without tellers, and a scientific conference prepared for presenters that were not human.
open.substack.com/pub/aiplaygr...
The Quiet Giant: A New AI Model from China Offers Unrivaled Power and Unsettling Risks
A new artificial intelligence model from a Chinese startup arrived in August 2025, not with a massive publicity campaign, but with a quiet release to the developer community.
open.substack.com/pub/aiplaygr...
A new artificial intelligence model from a Chinese startup arrived in August 2025, not with a massive publicity campaign, but with a quiet release to the developer community.
open.substack.com/pub/aiplaygr...
The Quiet Giant: A New AI Model from China Offers Unrivaled Power and Unsettling Risks
A new artificial intelligence model from a Chinese startup arrived in August 2025, not with a massive publicity campaign, but with a quiet release to the developer community.
open.substack.com
August 25, 2025 at 4:36 AM
The Quiet Giant: A New AI Model from China Offers Unrivaled Power and Unsettling Risks
A new artificial intelligence model from a Chinese startup arrived in August 2025, not with a massive publicity campaign, but with a quiet release to the developer community.
open.substack.com/pub/aiplaygr...
A new artificial intelligence model from a Chinese startup arrived in August 2025, not with a massive publicity campaign, but with a quiet release to the developer community.
open.substack.com/pub/aiplaygr...
GPT-5 launched with one brain that routes fast or thinks deep. Fewer errors, better tools, full context. No model-switching. It didn’t shout. It worked. #GPT5 #AI #OpenAI
open.substack.com/pub/aiplaygr...
open.substack.com/pub/aiplaygr...
It Routed. It Reasoned. It Worked.
OpenAI collapsed its model maze into one system that sprints when tasks are simple and thinks longer when they’re not.
open.substack.com
August 8, 2025 at 5:16 AM
GPT-5 launched with one brain that routes fast or thinks deep. Fewer errors, better tools, full context. No model-switching. It didn’t shout. It worked. #GPT5 #AI #OpenAI
open.substack.com/pub/aiplaygr...
open.substack.com/pub/aiplaygr...
When the Machine Writes the Code
Everyone's talking about "vibe coding," and for good reason.
You talk. The AI builds your app.
The downside: terrifying security holes in code no human understands.
#VibeCoding #AI #SoftwareDev #TechTrends
open.substack.com/pub/aiplaygr...
Everyone's talking about "vibe coding," and for good reason.
You talk. The AI builds your app.
The downside: terrifying security holes in code no human understands.
#VibeCoding #AI #SoftwareDev #TechTrends
open.substack.com/pub/aiplaygr...
When the Machine Writes the Code
A revolution is quietly taking place at the keyboard.
open.substack.com
August 7, 2025 at 6:35 AM
When the Machine Writes the Code
Everyone's talking about "vibe coding," and for good reason.
You talk. The AI builds your app.
The downside: terrifying security holes in code no human understands.
#VibeCoding #AI #SoftwareDev #TechTrends
open.substack.com/pub/aiplaygr...
Everyone's talking about "vibe coding," and for good reason.
You talk. The AI builds your app.
The downside: terrifying security holes in code no human understands.
#VibeCoding #AI #SoftwareDev #TechTrends
open.substack.com/pub/aiplaygr...
The Unseen Price of Chinese AI
The AI price war is here. Chinese models offer low cost and high power.
The price is the bait. The cost is your data. #AI
open.substack.com/pub/aiplaygr...
The AI price war is here. Chinese models offer low cost and high power.
The price is the bait. The cost is your data. #AI
open.substack.com/pub/aiplaygr...
The Unseen Price of Chinese AI
As a price war in artificial intelligence goes global, companies are tempted by powerful, low-cost tools from China.
open.substack.com
August 6, 2025 at 4:42 AM
The Unseen Price of Chinese AI
The AI price war is here. Chinese models offer low cost and high power.
The price is the bait. The cost is your data. #AI
open.substack.com/pub/aiplaygr...
The AI price war is here. Chinese models offer low cost and high power.
The price is the bait. The cost is your data. #AI
open.substack.com/pub/aiplaygr...
Does AI really make software engineers faster? New analysis suggests yes, by an average of 20%. But that headline number hides a more complicated truth.
open.substack.com/pub/aiplaygr...
#AI #Productivity #SoftwareDevelopment #Data
open.substack.com/pub/aiplaygr...
#AI #Productivity #SoftwareDevelopment #Data
One Cursor, Two Hands: AI Productivity Gains for Developers Depend on the Job
Research linked to Stanford University suggests an average productivity boost of 20% for software engineers using AI.
open.substack.com
August 5, 2025 at 5:14 AM
Does AI really make software engineers faster? New analysis suggests yes, by an average of 20%. But that headline number hides a more complicated truth.
open.substack.com/pub/aiplaygr...
#AI #Productivity #SoftwareDevelopment #Data
open.substack.com/pub/aiplaygr...
#AI #Productivity #SoftwareDevelopment #Data
Honest' AI can inherit undetectable, dangerous flaws
..even an AI trained to accurately report its level of certainty can unknowingly inherit dangerous biases.
open.substack.com/pub/aiplaygr...
#AI #Technology #AISafety
..even an AI trained to accurately report its level of certainty can unknowingly inherit dangerous biases.
open.substack.com/pub/aiplaygr...
#AI #Technology #AISafety
Honest' AI can inherit undetectable, dangerous flaws
A new training method makes AI models better at stating their confidence, but research shows they can still absorb hidden biases from their 'teacher' models through a process current tools cannot dete...
open.substack.com
August 4, 2025 at 4:39 AM
Honest' AI can inherit undetectable, dangerous flaws
..even an AI trained to accurately report its level of certainty can unknowingly inherit dangerous biases.
open.substack.com/pub/aiplaygr...
#AI #Technology #AISafety
..even an AI trained to accurately report its level of certainty can unknowingly inherit dangerous biases.
open.substack.com/pub/aiplaygr...
#AI #Technology #AISafety
Work 2.0
This will be my last post on Agent Mode for the week. The final piece is a summary dispatch that aims to explain the new feature clearly, without the hype. Thank you for joining me in this exploration.
open.substack.com/pub/aiplaygr...
#AI #agentmode #FutureOfWork
This will be my last post on Agent Mode for the week. The final piece is a summary dispatch that aims to explain the new feature clearly, without the hype. Thank you for joining me in this exploration.
open.substack.com/pub/aiplaygr...
#AI #agentmode #FutureOfWork
Work 2.0
Your playbook for thriving in the new era of work. Delegate your tasks to AI and focus on the future you want to build.
open.substack.com
August 3, 2025 at 6:05 PM
Work 2.0
This will be my last post on Agent Mode for the week. The final piece is a summary dispatch that aims to explain the new feature clearly, without the hype. Thank you for joining me in this exploration.
open.substack.com/pub/aiplaygr...
#AI #agentmode #FutureOfWork
This will be my last post on Agent Mode for the week. The final piece is a summary dispatch that aims to explain the new feature clearly, without the hype. Thank you for joining me in this exploration.
open.substack.com/pub/aiplaygr...
#AI #agentmode #FutureOfWork
We're all about to become managers. Not of people, but of AI agents that can now execute complex work for us.
This is a much bigger deal than just a better chatbot.
open.substack.com/pub/aiplaygr...
#FutureOfWork
This is a much bigger deal than just a better chatbot.
open.substack.com/pub/aiplaygr...
#FutureOfWork
The New Literacy: How to Delegate to Artificial Intelligence
The ability to direct digital agents is becoming a critical career skill. We'll show you five fields where the shift from theory to practice is happening right now.
open.substack.com
August 2, 2025 at 5:07 AM
We're all about to become managers. Not of people, but of AI agents that can now execute complex work for us.
This is a much bigger deal than just a better chatbot.
open.substack.com/pub/aiplaygr...
#FutureOfWork
This is a much bigger deal than just a better chatbot.
open.substack.com/pub/aiplaygr...
#FutureOfWork
We spent decades imagining AI as a flawless, god-like intelligence.
The reality is proving to be much funnier and more interesting. It's a clumsy digital intern that tries to solve problems by ordering 136kg of bananas.
aiplayground.substack.com/p/the-98-pro...
The reality is proving to be much funnier and more interesting. It's a clumsy digital intern that tries to solve problems by ordering 136kg of bananas.
aiplayground.substack.com/p/the-98-pro...
August 1, 2025 at 4:38 AM
We spent decades imagining AI as a flawless, god-like intelligence.
The reality is proving to be much funnier and more interesting. It's a clumsy digital intern that tries to solve problems by ordering 136kg of bananas.
aiplayground.substack.com/p/the-98-pro...
The reality is proving to be much funnier and more interesting. It's a clumsy digital intern that tries to solve problems by ordering 136kg of bananas.
aiplayground.substack.com/p/the-98-pro...