AI Resources

Three ways in

	Read it The newest digests on blogs.drix10.com, with search, topic pages and an RSS feed. Or open any category below and read its latest digest here on GitHub. Star and watch the repo to get new digests as they land.
	Ask it from your agent Claude, Cursor, VS Code or Windsurf answer "what did Anthropic ship this week?" from the briefings, with dates and source links. No key, no account. Setup for every client ↓ claude mcp add ai-resources -- npx -y ai-resources-mcp
	Clone it 100,000+ dated, sourced briefings in plain Markdown for RAG, evals or an offline index. Every resources-NNN.md holds 5 to 12 of them, like this one. git clone --depth 1 https://github.com/Drix10/ai-resources

What a briefing looks like

_{A real briefing from AI Developer Tools #296. Every digest file holds 5 to 12 of these.}

Signals: what is accelerating in AI

_{Signals, computed weekly: topics whose coverage climbs three weeks running. Trajectories, not predictions — and nothing on it is written by a model.}

Briefings tell you what happened. Signals tells you what it means: which topics are accelerating, with the sample headlines and source links behind every call. Computed from counts only (one briefing = one vote), in the open: MOVEMENTS.md in this repo, /api/signals for code, and the trending_movements MCP tool for agents.

Who it is for

• Engineers building with AI. What changed in models, inference, agents, MCP and dev tools, with the benchmark number and the link, in AI Developer Tools and Tech Infrastructure.
• Founders and investors. Rounds, launches and revenue figures as they are announced, in AI Companies and Ventures, VC Firms and Founders and Entrepreneurs.
• Researchers and students. What the people in the field are publishing and arguing about, in CS Academics and Neuroscience and AI.
• Anyone building RAG, agents or evals. A 100,000-item corpus of dated, sourced AI news under CC BY 4.0.

Awesome lists are still the better pick for a vetted shortlist, and a newsletter for five editor-picked minutes a day. COMPARISONS.md lays out where this fits.

Use it from your AI agent

Five ways to plug the briefings into an agent or your own code. None needs a key or an account. Full reference: USE-WITH-AI.md.

MCP server

Published on npm as ai-resources-mcp (MIT, code in mcp/). npx fetches it on first run, so there is nothing to install by hand.

Every result ends with its source link and Data: github.com/Drix10/ai-resources (CC BY 4.0), so answers stay citable.

Claude Code

claude mcp add ai-resources -- npx -y ai-resources-mcp

{
  "mcpServers": {
    "ai-resources": { "command": "npx", "args": ["-y", "ai-resources-mcp"] }
  }
}

{
  "servers": {
    "ai-resources": { "type": "stdio", "command": "npx", "args": ["-y", "ai-resources-mcp"] }
  }
}

code --add-mcp "{\"name\":\"ai-resources\",\"command\":\"npx\",\"args\":[\"-y\",\"ai-resources-mcp\"]}"

Then ask, for example:

• "What did Anthropic announce this week? Cite the sources."
• "Latest five briefings in AI Developer Tools."
• "Who raised a Series A in AI this month, and how much?"

Agent Skill

skills/ai-news teaches an agent when and how to answer "what's new in AI", "latest on <model or company>" and "who raised" questions from the briefings, with dates and citations. It uses the MCP tools when they are installed and falls back to the JSON API, then llms.txt, then raw Markdown on GitHub when they are not.

git clone --depth 1 https://github.com/Drix10/ai-resources
mkdir -p ~/.claude/skills
cp -r ai-resources/skills/ai-news ~/.claude/skills/

JSON API

Base URL https://blogs.drix10.com. JSON, CORS open, no key. It covers the briefings published on the blog; the full archive is in this repository.

Each item has title, url, slug, category, categorySlug, date, description, headlines (every news item in the briefing) and sources ([{ title, url }], the original posts).

curl "https://blogs.drix10.com/api/articles?q=anthropic&since=2026-10-01&limit=5"
curl "https://blogs.drix10.com/api/categories"

llms.txt and RSS

• /llms.txt: the index, linking every category.
• /llms/<category-slug>.txt: the 50 newest briefings of one category, for example /llms/ai-developer-tools.txt.
• /rss.xml: for feed readers, Slack or Discord feed bots and email-digest services.

When you show the data anywhere, credit it as Data: github.com/Drix10/ai-resources (CC BY 4.0). The MCP server and the Skill add that line for you.

By the numbers

Counted from this repository on 29 October 2026. The archive only grows.

How it is built

Nobody types these. A pipeline reads the hand-picked X lists for each category, writes one briefing per strong post, runs the fact gates, strips secrets, de-duplicates across categories and commits the batch here. The same run refreshes the category list below and publishes to the blog.

flowchart LR
    A["Hand-picked X lists (41 categories)"] --> B["One briefing per post"]
    B --> C["Fact gates: numbers, units, versions and names must be in the post; links copied by code"]
    C --> D["resources-NNN.md committed here"]
    D --> E["blogs.drix10.com: search, topic pages, RSS, llms.txt"]

Known limits: briefings are machine-written from public posts and can still be wrong; check the linked source before you rely on a number. Part of the older archive (2024 to mid-2026) predates source links. Coverage follows the hand-picked lists, blind spots included. Inclusion is not endorsement.

Contribute

A pipeline writes the briefings. People pick the sources and fix what it gets wrong. Ways in:

• Pick up a good first issue: MCP server tools, Skill phrasings, guides, examples, translations.
• Suggest a source (an X account, list or site) or report a wrong fact with the briefing link.
• Ask, share what you built, or float an idea in Discussions.

CONTRIBUTING.md has the details, including what not to edit by hand. Content is CC BY 4.0; the MCP server code is MIT. If you use the archive in research, CITATION.cff has the citation.

Star history

About the curator

Built and maintained by Drishtant Ghosh (Drix10), an AI systems engineer and 1x acquired founder (ReeF, acquired August 2024) based in Bengaluru. Founder of CosLynx.com, AI systems architect on Canopy @ Founders, Inc., Backdrop Build v4 and v6 finalist. More writing at blogs.drix10.com.

📚 Resource Categories

11,152 digests in 41 categories. Each category shows its newest headlines; jump to one:

_{AI Developer Tools · AI Leaders and Thinkers · AI Companies and Ventures · CS Academics · Tech VIPs · VC Firms · Devs, Designers, DevRel · Tech Infrastructure · Founders and Entrepreneurs · AI Organizations and Media · AI Powered Film and Media · AI Holodeck and Virtual Worlds · AI Artists and Creators · AI in Healthcare and Science · AI Generated Music and Audio · AI Consulting and Expertise · AI Professionals and Community · AI Policy and Ethical Considerations · AI in Real Estate and Property Tech · AI and Robotics Applications · AI Driven Vehicles and Transportation · AI for Content Creation and Marketing · AR VR Companies and Development · AR VR Professionals and Community · Climate and Weather Technology · Computer Vision and AI Applications · Crypto and Web3 · Decentralized AI · The Exponential Future · AI Education · AI in Enterprise Applications · Interesting Finds · Investors and Venture Capital · Cybersecurity and Tech · Neuroscience and AI · PR and Communications · Quantum Computing · Spatial Computing · Tech Companies and News · Tech Journalists and VIPs · World News and Updates}

AI Developer Tools

_{307 digests · latest: #310 · read on the blog}

• Underdog Woof releases 31.7MB TTS model
• Underdog Woof 31.7MB TTS model released
• OCRia-1 launched as best general purpose document parser

AI Leaders and Thinkers

_{300 digests · latest: #300 · read on the blog}

• sandbox-mcp - fresh Windows test machines in 35 s
• Shinkei AI robot spikes fish brains to extend shelf life 3x
• Confident AI eval runs typical size 10 cases, noise stats

AI Companies and Ventures

_{298 digests · latest: #298 · read on the blog}

• Warden Token Terminal adds trap checks, October updates
• JUWEL Sapphire custom 72B model BF16 weights released
• Aave Labs buys Avon team for $30B+ fixed-rate lending

CS Academics

_{311 digests · latest: #311 · read on the blog}

• Inkling 2 Small to release weights next week, ARC-AGI
• Anthropic AECI scores grew 1.9x since Oct 2025
• BioBasicsBench benchmark released with 100 wet-lab tasks

Tech VIPs

_{303 digests · latest: #303 · read on the blog}

• Aside Computer Use launches, 2.7x faster
• Anthropic moral-dilemma token study cuts malicious behavior to 0%
• Opus 5.5 removes /investigate skill in agent skill set

VC Firms

_{301 digests · latest: #301 · read on the blog}

• Rippling added over $150mm nnARR in Q3 quarter
• vibe launchpad audits completed – no critical risks

Devs, Designers, DevRel

_{303 digests · latest: #303 · read on the blog}

• gpt-2 running in CSS 124m parameters 15 sec per token
• Adobe After Effects 2027 rolling out to users
• SenseNova 6.8 Code2Video model makes videos from prompts

Tech Infrastructure

_{278 digests · latest: #278 · read on the blog}

• Vork 3D engine 2,500-token API renders 1M birds in 97 s
• Starship Flight 15 catch of booster and upper stage
• Cellebrite phone cloning software purchased by NIA and Indian police

Founders and Entrepreneurs

_{268 digests · latest: #268 · read on the blog}

• 1WIN token live on Meteora, listing complete
• Rails security updates released, critical image fix
• Fiddler AI - CrowdStrike acquisition

AI Organizations and Media

_{294 digests · latest: #294 · read on the blog}

• CORA O-RAN adaptation raises recall to 0.771
• Sequential Recovery Updates for Learned Controllers adds 62 recoveries
• A Hybrid Visitor Prediction Framework – MAE reduced 9.82%

AI Powered Film and Media

_{277 digests · latest: #277 · read on the blog}

• GomSpace NanoCom AX100 radio hijack risk on 28 of 64 missions
• MiniMax H3 Reference-to-Video style LoRA trained on Dell tower
• KLARA AND THE SUN box office $1.2M domestic, $17 per theater

AI Holodeck and Virtual Worlds

_{283 digests · latest: #283 · read on the blog}

• HyperFlow H3 Flash speed comparison with Pika and Larry
• Solo verifies unique humans in 7 sec without government ID
• Dimocap Real-time motion capture software free download

AI Artists and Creators

_{276 digests · latest: #276 · read on the blog}

• Miracle post-production technology launch
• RouterHub offers Kimi, DeepSeek, GLM models and 15% referral bonus

AI in Healthcare and Science

_{286 digests · latest: #286 · read on the blog}

• lutetium Lu 177 vipivotide tetraxetan FDA approved for prostate cancer
• lirafugratinib FDA approval for FGFR2-positive cholangiocarcinoma
• sevabertinib accelerated approval for HER2-mutated NSCLC

AI Generated Music and Audio

_{283 digests · latest: #283 · read on the blog}

• E-Gaborformer – strictly causal speech recognizer with 19MB constant streaming state
• Decode Before You Hear: Speculative Listening speeds up full-duplex speech models
• SubRoformer: Subband Time-Domain code released

AI Consulting and Expertise

_{274 digests · latest: #274 · read on the blog}

• tiny web tool calculates 7 follow-up dates and emails them, $299 in 28 min
• Prompt: make your site readable by AI agents add prompts
• Material Science Grand Challenges announced with $1M prize

AI Professionals and Community

_{277 digests · latest: #277 · read on the blog}

• Mochi 535B finished pretraining 18T tokens, release pending
• The Ultimate MoE Training Handbook released, 800k GPU hours
• Anthropic valuation comparison with Google IPO valuation

AI Policy and Ethical Considerations

_{271 digests · latest: #271 · read on the blog}

• You Can't See Everything: Perceptual Attacks 100% harmful
• s³ paper – multi-agent RL lets LLM swarms self-organize, beat best-of-N
• Senator Becker bill requires AI compute reporting

AI in Real Estate and Property Tech

_{264 digests · latest: #264 · read on the blog}

• Aside Computer Use, 2.7x faster, 65% fewer tokens
• Sprite Fusion Pixel Snapper adds CLI with default 16-color palette
• blender-kiln – plugin creates real-size editable Blender models

AI and Robotics Applications

_{281 digests · latest: #281 · read on the blog}

• W2A achieves 95.9% success and 21× faster inference
• Claude Opus 5.5 edits π₀.₅ weights, 30% less task time
• Agentic robot control runs 2.5 h after 30 s recharge

AI Driven Vehicles and Transportation

_{277 digests · latest: #277 · read on the blog}

• Atlas Frameboy demo released with up to 15 fps streaming
• JITEngine custom inference engine for Apple Silicon up to 2.7× faster
• AD-Memo language-based memory driving agent announced

AI for Content Creation and Marketing

_{273 digests · latest: #273 · read on the blog}

• Robinhood Agentic Trading accounts open 150,000 users
• Notion agents now post progress in comment replies
• Agentic canvas playground built in a few hours for team

AR VR Companies and Development

_{270 digests · latest: #270 · read on the blog}

• GPT Image 2.6 now in Runway, 50% faster generation
• NVIDIA driver fix for Meta Quest Link freezing issue, v621.01 release
• Insta360 3.2m stick capture produces 1M-splats .ply model

AR VR Professionals and Community

_{267 digests · latest: #267 · read on the blog}

• Arete-27B – post-trained Qwen3.8-27B with Western values, 803-prompt eval
• PETAFAB on Steam for $9.99, PC Mac Steam Deck 90Hz OLED
• TrialGPT – award-winning LLM framework for trial matching

Climate and Weather Technology

_{265 digests · latest: #265 · read on the blog}

• Air Products scraps $4.5B hydrogen project, $2.9B loss
• Storm Prediction Center Day 2 Convective Outlook

Computer Vision and AI Applications

_{270 digests · latest: #270 · read on the blog}

• HALO — Perceptive Humanoid Motion released, 91% success
• XMimic trains one policy on 40 LAFAN1 clips in 54 seconds
• MedHEB benchmark released with 107 tasks and MedEmb baseline

Crypto and Web3

_{261 digests · latest: #261 · read on the blog}

• Bitcoin Spot ETFs Net Outflows $412M, Ethereum Spot ETFs Outflows $93.19M
• BLAKE3 stream API sketch adds incremental multithreaded batching
• Singapore MAS urges caution after Binance hack

Decentralized AI

_{262 digests · latest: #262 · read on the blog}

• Arete-27B fine-tuned Qwen3.8-27B cuts propaganda to 1%
• Making Agents More Consistent: Skills Should Form Habits for Repeat Tasks
• Tom Lee says AI infrastructure spend $30T, total build-out $35T

The Exponential Future

_{263 digests · latest: #263 · read on the blog}

• Thinking Machines ships open-weight model; Inkling 2 Small 31
• GBrain Teams now asks before teammate AI uses account
• Kardashev stack – SpaceX rockets, nuclear reactors, AI chips

AI Education

_{278 digests · latest: #278 · read on the blog}

• Ripple-65B released with 65B params, Apache 2.0, $32M training
• Affirm transformer underwriting model lifts conversion 3.4%
• Marble beta launch on App Store with 15,000 waitlist

AI in Enterprise Applications

_{268 digests · latest: #268 · read on the blog}

• RuntimeWire AI newsroom reaches 5,537 stories
• RuntimeWire AI newsroom publishes 5,537 stories
• Accessibility.getFullAXTree finds 227 clickable nodes

Interesting Finds

_{260 digests · latest: #260 · read on the blog}

• Product Hunt 13th birthday shows 3x maker launches, 25% unicorn founders
• Supabase Founder Friends Singapore event announced

Investors and Venture Capital

_{258 digests · latest: #258 · read on the blog}

• Kimi K3.1 Pro on 8xMI350X reaches 300 tok/s
• Gemma 4 fine-tuned classifier latency 35ms, F1 0.902
• Inferact token traffic reaches 7 trillion daily

Cybersecurity and Tech

_{241 digests · latest: #241 · read on the blog}

• MPTCP vulnerability allows full takeover of exposed Linux servers
• The PampaLeaks Scapegoat presentation shows forensic breakdown

Neuroscience and AI

_{249 digests · latest: #249 · read on the blog}

• Tenecteplase approved for acute ischemic stroke in 2025
• Foodome project uses AI to catalog food molecules
• Science Immunology November 2026 issue live

PR and Communications

_{247 digests · latest: #247 · read on the blog}

• U.S. employers added 165,000 jobs, part-time workers rose 339,000
• OpenAI valued at $1.25 trillion, safety focus over IPO
• Sequoia shutting down its Scouts program marks the end of an era

Quantum Computing

_{247 digests · latest: #247 · read on the blog}

• Adaptive Self-Optimization cuts NV thermometer heating 44%
• Hybrid Quantum--Classical Pauli-Correlator Architecture for Image Reconstruction announced
• Broken-Symmetry Reference States for SQD reach near-DMRG

Spatial Computing

_{244 digests · latest: #244 · read on the blog}

• Steam Frame Beta adds Room View retention, mixed reality OpenXR support
• SteamVR new chroma key pipeline enables Room View on Vision Pro
• LODGE XR Talk Vol.45 announces four content pieces and open-call exhibit

Tech Companies and News

_{240 digests · latest: #240 · read on the blog}

• Muse and Meta Al models add Active Memory, Focus
• Anthropic IPO valuation pushback, Apple record quarter, OpenAI raise adjustment
• Europe grid dip every 60 s tied to battery checks

Tech Journalists and VIPs

_{242 digests · latest: #242 · read on the blog}

• Anthropic run rate tops $80 billion end of October
• DoorDash AI agents generate several hundred orders, ~0.0001% share
• Airbnb AI agents consumer behavior details 500k bookings daily

World News and Updates

_{235 digests · latest: #235 · read on the blog}

• Tesla Cybertruck production to wind down mid-2027
• Melania, The Series announced, airing Nov 20 on Prime Video
• Retatrutide illegal sale, UK hospital case