Wiredframe Radar logo

Category intelligence

Social Media Briefing — January 13, 2026

446 current items analyzed and ranked.

Executive synthesis

Social Media Summary

Two major stories dominated AI discussions: the Apple-Google AI partnership and Anthropic's Cowork launch.

  • Jeff Dean announced Gemini models will power Apple Intelligence, with Apple calling Google's AI "the most capable foundation" for their needs
  • Anthropic launched Cowork, a general-purpose agent for non-coding tasks like vacation research and email management—garnering 6.9K likes and 724K views
  • Simon Willison published early impressions of Cowork, noting the $100+/month pricing tier

Healthcare AI saw aggressive expansion: OpenAI acquired Torch (1.1M views) while Anthropic announced new healthcare connectors and Agent Skills.

Ethical concerns about AI companies scraping public content "for free, then selling it back as tokens" sparked heated debate.

Key Themes

Anthropic Products & Claude Ecosystem · 12Apple-Google AI Partnership · 35Google-Apple AI Partnership · 4Healthcare AI Expansion · 5Gemini API Updates · 5Apple-Google Partnership & OpenAI Strategy · 8Transformer Architecture Research · 3AI Agent Development · 8Positional Encoding Breakthroughs · 3Open Source Model Ecosystem · 5

Primary evidence

Top Ranked Signals

Social Twitter Jan 12

Since we launched Claude Code, we saw people using it for all sorts of non-coding work: doing vacati...

By @bcherny

95 score
AI Analysis

Major Anthropic product announcement: Introducing 'Cowork', an AI agent for non-coding tasks like vacation research, slide decks, email management. Features include built-in VM isolation, browser automation, and data connectors. Available as research preview for Claude Max subscribers on macOS.

Since we launched Claude Code, we saw people using it for all sorts of non-coding work: doing vacation research, building slide decks, cleaning up your email, cancelling subscriptions, recovering wedding photos from a hard drive, monitoring plant growth, controlling your oven. These use cases are diverse and surprising -- the reason is that the underlying Claude Agent is the best agent, and Opus 4.5 is the best model. Today, we're so excited to introduce Cowork, our first step towards making
anthropic_productsai_agentscomputer_useproduct_launch
Social Twitter Jan 12

“After careful evaluation, Apple determined that Google's Al technology provides the most capable fo...

By @OfficialLoganK

95 score
AI Analysis

Google employee sharing news that Apple has chosen Google's AI technology as the foundation for Apple Foundation Models, signaling a major partnership

“After careful evaluation, Apple determined that Google's Al technology provides the most capable foundation for Apple Foundation Models and is excited about the innovative new experiences it will unlock for Apple users.” : )
Industry PartnershipsApple AIGoogle AI
Social Twitter Jan 12

One of my favorite findings: Positional embeddings are just training wheels. They help convergence b...

By @hardmaru

92 score
AI Analysis

David Ha shares major finding: positional embeddings help convergence but hurt long-context generalization; deleting them after pretraining and recalibrating for <1% budget unlocks massive context windows

One of my favorite findings: Positional embeddings are just training wheels. They help convergence but hurt long-context generalization. We found that if you simply delete them after pretraining and recalibrate for < 1% of the original budget, you unlock massive context windows.
technical-researchpositional-embeddingstransformerscontext-lengthmodel-architecture
Social Twitter Jan 12

Here's more analysis on the Apple and Google deal to make a new kind of Siri, after I had a cup of c...

By @Scobleizer

92 score
AI Analysis

Following yesterday's News coverage, Comprehensive analysis of Apple-Google Siri deal, OpenAI becoming products company, AR glasses race, multimodal AI for robotics, patent landscape, and Apple's retail/content advantages

Here's more analysis on the Apple and Google deal to make a new kind of Siri, after I had a cup of coffee. This is what OpenAI is doing: they're making a variety of new products and going after Apple. Apple didn't want to give OpenAI any more data to help a potential new competitor. The real problem for this OpenAI effort is that we're about to move to glasses. People don't believe me that we're about to move to glasses, but you should, because I just got back from CES and there was a ton of g
apple_google_partnershipAR_glassesopenai_strategymultimodal_airoboticsai_patentswearables
Social Twitter Jan 12

We’ve acquired Torch, a healthcare startup that unifies lab results, medications, and visit recordin...

By @OpenAI

90 score
AI Analysis

OpenAI announces acquisition of Torch, a healthcare startup unifying lab results, medications, and visit recordings, to enhance ChatGPT Health

We’ve acquired Torch, a healthcare startup that unifies lab results, medications, and visit recordings. Bringing this together with ChatGPT Health opens up a new way to understand and manage your health. We're excited to welcome the Torch team to OpenAI @IlyaAbyzov, @elh_online, @jfhamlin, and Ryan Oman.
openaihealthcare-aiacquisitionschatgptproduct-expansion
Social Twitter Jan 12

Excited to announce the Relative Adoption Metric a new way of studying model downloads that contextu...

By @natolambert

88 score
AI Analysis

Nathan Lambert announces the 'Relative Adoption Metric' (RAM Score), a new methodology for studying model downloads that accounts for size-category biases. Analysis shows GPT-OSS has exceptional adoption, MiniMax/Moonshot performing well in large MoE space.

Excited to announce the Relative Adoption Metric a new way of studying model downloads that contextualizes it across time and model sizes. While building The ATOM Project and other tools to measure the open ecosystem at @interconnectsai, we are often frustrated with using downloads as a primary metric. We, and the community, know that small models are downloaded much more, so it makes some adoption metrics favor organizations releasing small models. Over the 1,100+ leading LLMs we track careful
open_source_modelsai_ecosystem_analysismodel_adoptionresearch_methodology
Social Twitter Jan 12

Today we are shipping expanded URL support 🔗 in the Gemini API!! You can now pass signed or public U...

By @OfficialLoganK

88 score
AI Analysis

Google announces expanded URL support in Gemini API, allowing direct passing of signed/public URLs for images/PDFs, plus Google Cloud Storage integration

Today we are shipping expanded URL support 🔗 in the Gemini API!! You can now pass signed or public URLs directly to Gemini and it will fetch images or PDFs, along with support for a new integration with Google Cloud Storage allowing you to keep you data in one place!!
Gemini APIDeveloper ToolsAPI Features
Social Bluesky Jan 12

Introducing DroPE: Extending Context by Dropping Positional Embeddings We found embeddings like RoP...

By @sakanaai.bsky.social

88 score
AI Analysis

Introducing DroPE: extending context by dropping positional embeddings after training; RoPE aids training but bottlenecks long-sequence generalization

Introducing DroPE: Extending Context by Dropping Positional Embeddings We found embeddings like RoPE aid training but bottleneck long-sequence generalization. Our solution’s simple: treat them as a temporary training scaffold, not a permanent necessity. arxiv.org/abs/2512.12167 pub.sakana.ai/DroPE
positional_encodingtransformerslong_contextllm_architectureresearch
Social Bluesky Jan 12

One of my favorite findings: Positional embeddings are just training wheels. They help convergence b...

By @hardmaru.bsky.social

85 score
AI Analysis

Key finding: positional embeddings are 'training wheels' - help convergence but hurt long-context; deleting after pretraining and recalibrating unlocks massive context windows

One of my favorite findings: Positional embeddings are just training wheels. They help convergence but hurt long-context generalization. We found that if you simply delete them after pretraining and recalibrate for <1% of the original budget, you unlock massive context windows. Smarter, not harder.
positional_encodingtransformerslong_contextllm_architecture
Social Twitter Jan 12

To support the work of the healthcare and life sciences industries, we're adding over a dozen new co...

By @AnthropicAI

82 score
AI Analysis

Anthropic announces new healthcare and life sciences connectors and Agent Skills for Claude, with livestream to discuss implementation

To support the work of the healthcare and life sciences industries, we're adding over a dozen new connectors and Agent Skills to Claude. We're hosting a livestream at 11:30am PT today to discuss how to use these tools most effectively. Learn more: t.co/SsKZFaZlnQ
anthropichealthcare-aiai-agentsenterpriseproduct-launch
Social Twitter Jan 12

We're feeding AI our best work for free, and nobody is talking about what happens next. AI will scr...

By @svpino

82 score
AI Analysis

Commentary on AI companies scraping public content (blogs, code, tutorials) without compensation, then selling it back as tokens. Warns that content creation economics are broken and need fixing.

We're feeding AI our best work for free, and nobody is talking about what happens next. AI will scrape every blog and social media post you publish. AI will scrape every single open-source code you share. AI will scrape every tutorial you record. And then, they will sell this info back to you in the form of tokens and soon, ads. The economics of the future of content creation are broken. If we don't fix this, we will regret it.
ai_ethicstraining_datacontent_economicsdata_scraping