Wiredframe Radar logo

Category intelligence

Social Media Briefing — January 8, 2026

493 current items analyzed and ranked.

Executive synthesis

Social Media Summary

OpenAI dominated discussions with the ChatGPT Health launch, integrating medical records and wellness apps for their 230M+ weekly health questioners. A buried bombshell: Greg Brockman casually revealed using GPT-5.2 for solving an open Erdős problem.

  • Andrej Karpathy released nanochat miniseries v1, reproducing Chinchilla scaling laws with deeply technical methodology
  • Andrew Wilson (NYU) introduced epiplexity, a novel information measure defining what bounded observers can extract from data
  • Jeremy Howard shared ironic proof of llms.txt value: Tailwind rejected adding it specifically because it would be too useful

The Claude Code conversation continued with Ethan Mollick demonstrating building businesses and simulations as a non-coder. Matt Shumer sought founders building 'Slack for agents,' signaling serious investor appetite for multi-agent infrastructure. Andrew Ng launched a free course teaching vibe coding to non-programmers, further democratizing AI development.

Key Themes

LLM Scaling & Training Efficiency · 4ChatGPT Health Launch · 3ChatGPT Health & AI in Healthcare · 8AI Coding Assistants & Claude Code · 6Novel Research & Information Theory · 10Practical AI Applications · 4llms.txt Adoption and Value · 4AI Coding Agents & Claude Code · 2AI Democratization & No-Code Development · 5AI Business Strategy & Valuations · 3

Primary evidence

Top Ranked Signals

Social Twitter Jan 7

New post: nanochat miniseries v1 The correct way to think about LLMs is that you are not optimizing...

By @karpathy

95 score
AI Analysis
Karpathy releases nanochat miniseries v1, demonstrating LLM scaling laws that reproduce Chinchilla paper results, showing compute-optimal training can match GPT-2 for ~$500 (targeting <$100)
New post: nanochat miniseries v1 The correct way to think about LLMs is that you are not optimizing for a single specific model but for a family models controlled by a single dial (the compute you wish to spend) to achieve monotonically better results. This allows you to do careful science of scaling laws and ultimately this is what gives you the confidence that when you pay for "the big run", the extrapolation will work and your money will be well spent. For the first public release of nanocha
LLM scaling lawscompute optimizationopen source AI researchtraining efficiency
Social Twitter Jan 7

Introducing ChatGPT Health — a dedicated space for health conversations in ChatGPT. You can securely...

By @OpenAI

92 score
AI Analysis
OpenAI officially introduces ChatGPT Health as dedicated health conversation space with medical records and wellness app integration
Introducing ChatGPT Health — a dedicated space for health conversations in ChatGPT. You can securely connect medical records and wellness apps so responses are grounded in your own health information. Designed to help you navigate medical care, not replace it. Join the waitlist to get early access. t.co/MdpqDg7Ecg
AI healthproduct launchesChatGPT Healthmedical AI
Social Twitter Jan 7

✨ I used Claude Code to one-shot a 💁‍♀️ Karen Bot When we walk or drive around we come across a lot...

By @levelsio

92 score
AI Analysis
Levelsio built a 'Karen Bot' using Claude Code that allows him and his girlfriend to report local infrastructure problems (potholes, damaged bins, missing mirrors) to their city council. The app takes a photo, description, and map location, then ChatGPT writes a formal Portuguese letter with GPS coordinates and sends it automatically.
✨ I used Claude Code to one-shot a 💁‍♀️ Karen Bot When we walk or drive around we come across a lot of things here that could be improved in terms of public infrastructure Roads near us have lots of potholes, there's municipal waste bins here that someone put on fire 6 months ago, and they're all melted away and still not replaced, and the roadside mirror at a dangerous crossing suddenly disappeared so now when you wanna turn in there you might get because so many blindspots When we emailed
AI coding assistantspractical AI applicationsClaude Codecivic techautomation
Social Twitter Jan 7

🚨 OpenAI just launched ChatGPT Health. This is for all of you who have asked ChatGPT to explain tha...

By @alliekmiller

92 score
AI Analysis
Allie K Miller provides comprehensive overview of OpenAI's ChatGPT Health launch - connects to medical records, Apple Health, MyFitnessPal, Peloton; can explain labs, prep doctor questions, compare insurance; built with 260+ physicians over 2 years
🚨 OpenAI just launched ChatGPT Health. This is for all of you who have asked ChatGPT to explain that weird rash. Or that lingering cough. Or that itchy ear. If you've been paying attention, this release isn't a surprise. @OpenAI spent all week running health-focused ads - a mom using AI to squeeze workouts into her schedule, a man using it to understand his health conditions, a woman checking if fabrics will trigger her eczema while shopping. The writing was on the wall. And if you read my 20
ChatGPT HealthAI in healthcareOpenAIProduct launchMedical AI
Social Twitter Jan 7

How useful is llms.txt? It's so useful that Tailwind rejected a PR to add an llms.txt, on the basis...

By @jeremyphoward

90 score
AI Analysis
Jeremy Howard highlights ironic proof of llms.txt usefulness: Tailwind rejected PR to add llms.txt specifically because it would be so useful people wouldn't need their docs
How useful is llms.txt? It's so useful that Tailwind rejected a PR to add an llms.txt, on the basis that it would be so useful that people wouldn't need to read their docs any more! t.co/fn8Co36rmC t.co/68U8dOrzYF
llms.txtAI documentationDeveloper toolsLLM integration
Social Twitter Jan 7

We introduce epiplexity, a new measure of information that provides a foundation for how to select, ...

By @andrewgwils

88 score
AI Analysis
As first reported in Research yesterday Andrew Wilson introduces epiplexity, a new information measure for data selection, generation, and transformation in learning systems
We introduce epiplexity, a new measure of information that provides a foundation for how to select, generate, or transform data for learning systems. We have been working on this for almost 2 years, and I cannot contain my excitement! 1/7
epiplexityinformation theorydata selectionlearning theorynovel research
Social Twitter Jan 7

Is anyone building a “Slack for agents”? If so, I’d love to fund you + blow you up with my platform...

By @mattshumer_

88 score
AI Analysis
Matt Shumer seeking founders building 'Slack for agents' - multi-agent communication platform, offering funding and growth support, mentions backing Groq and Rork
Is anyone building a “Slack for agents”? If so, I’d love to fund you + blow you up with my platform. I’ve kickstarted growth for companies like @GroqInc, @rork_app and more. Would love to do this again for the right company. If this is you, reach out.
Multi-agent systemsAI infrastructureVenture investmentAgent coordination
Social Twitter Jan 7

If you’ve never written code before, this is for you. I’ve just launched a course that shows you, in...

By @AndrewYNg

85 score
AI Analysis
Andrew Ng launches free 30-minute course teaching non-coders to build web apps using AI, emphasizing 'vibe coding' and vendor-neutral approach
If you’ve never written code before, this is for you. I’ve just launched a course that shows you, in less than 30 minutes, how to describe an idea for an app and build it with AI. In this course, you'll build a working web application - a funny interactive birthday message generator that runs in your browser and can be shared with friends. You'll customize it by telling AI how you want it changed, and tweak it until it works the way you want. By the end, you'll have a repeatable process you can
AI educationvibe codingAI democratizationno-code development
Social Twitter Jan 7

that oai failed to turn ChatGPT's 900m weekly users into any form of lasting social app is probably ...

By @swyx

85 score
AI Analysis
Swyx argues that OpenAI's biggest consumer AI miss of 2025 was failing to turn ChatGPT's 900M weekly users into a lasting social app. He calculates that if each AI user was worth $5,750 (based on X/Grok valuations), OpenAI could potentially be worth $5T instead of $750B.
that oai failed to turn ChatGPT's 900m weekly users into any form of lasting social app is probably the biggest consumer ai miss of 2025* you can argue that oai did just fine ($157B -> $750B) NOT doing this, but you don't know the althistory where OAI suddenly became a full social network. X is $230B on 600m X MAU / 40m Grok MAU, in a world where every AI user is worth $5750 then the best-case OAI valuation would be around $5T rn. *yes i do like sora but no it's not a serious social network ye
AI business strategyOpenAIsocial networkingAI valuationsconsumer AI
Social Twitter Jan 7

I wrote about Claude Code and why non-coders should be paying attention (and playing with the system...

By @emollick

82 score
AI Analysis
Ethan Mollick writes about Claude Code's potential for non-coders, demonstrating how he used it to launch a business and build a civilization simulation game
I wrote about Claude Code and why non-coders should be paying attention (and playing with the system) - it shows what today’s LLMs can do Along the way I had Claude launch a business for me & build a game that simulates the rise and fall of civilizations. t.co/Pg3TQamYl9
AI democratizationcoding assistantspractical AI applications
Social Twitter Jan 7

I feel I'm operating at a much higher level when using Claude Code. I've heard some people saying t...

By @svpino

82 score
AI Analysis
Santiago (@svpino) reflects on productivity gains with Claude Code - argues AI doesn't make developers lazy but shifts work to higher-level thinking, specifications, and code review instead of writing boilerplate
I feel I'm operating at a much higher level when using Claude Code. I've heard some people saying that AI makes developers lazy, but that's not how I feel. I actually think I'm doing more work than ever before. Of course, I'm writing fewer lines of code myself, but who cares? Most of this code I wrote in my sleep. It's boilerplate code that I've written a million times before, and now I have a little genie that does it for me. The time I saved now went to reviewing Claude's code and working o
AI-assisted codingDeveloper productivityClaude CodeFuture of software development