More Perfect Union demonstrates that substance and distribution discipline can outperform trend-driven social tactics. Its YouTube-first model, now at nearly 3M subscribers with over half added in the past year, combines high-volume long-form production with a Shorts program that generates comparable viewership. Instagram carousels tied to live news moments accelerate follower growth, and video cut-downs are structured as complete stories rather than traffic funnels. The team publishes natively instead of relying on outbound links and invests heavily in titles and thumbnails to compete in crowded feeds. The core differentiator is pairing sharp framing with actionable solutions, positioning content to inform and mobilize rather than simply provoke reaction.
Google's WebMCP, which provides two APIs (Declarative, Imperative) that allow users to "mark up" where on your webpages AI-agents should be taking actions, is now in early access.
Anthropic's Super Bowl push drove Claude from 41st to 7th in the US App Store, lifting downloads 32% to 148,000 in 3 days as its ad-free positioning contrasted with ChatGPT's.
to power authentication, identity, and security. It's the infrastructure layer that makes applications enterprise ready without slowing teams down.
Sponsored
AI companies are scaling faster than ever and hitting enterprise requirements far earlier in their lifecycle.
Anthropic has closed a $30 billion funding round at a $380 billion post-money valuation. This is the second-largest private tech fundraising round on record after OpenAI, which closed a round of more than $40 billion last year. Anthropic's funding round was led by Coatue and Singapore sovereign wealth fund GIC. The fresh capital will support the company's infrastructure expansion, research, and continued investment in enterprise-grade products.
Waymo has begun using its sixth-generation driverless system to provide robotaxi rides to employees. The system uses more cost-effective parts and should be able to navigate through harsher weather conditions than previous generations. It will serve as the primary engine for Waymo's next era of expansion. The company is currently planning to extend its lead in the US while testing and planning for commercial operations abroad.
Europe's Ariane 6 rocket lifted off from the Guiana Space Center in South America at 11:45 AM EST on Thursday. The rocket's four boosters generated more than 3.4 million pounds of thrust as it hauled 32 spacecraft to low-Earth orbit for Amazon's satellite broadband constellation. The successful test demonstrates the European Space Agency's ability to deliver the most demanding large-scale constellation missions. The next Ariane 6 launch, to be announced in the coming weeks, will carry another set of Amazon Leo satellites to orbit.
2026.02.13·arstechnica.com·6 min read·TLDR Alphabet has updated its Gemini Deep Think AI model for better performance in math and science research. The model is now able to help scientists move from theoretical reasoning to practical applications. It uses Google Search to avoid inaccuracies and wrongful citations when doing research. Google has also developed a math research agent called Aletheia that can conduct autonomous research or collaborate with humans. Aletheia is available in the Gemini app for Google AI Ultra subscribers and also for select researchers.
2026.02.13·links.tldrnewsletter.com·2 min read·TLDR As the first engineer on TLDR's new Applied AI team, you'll build AI agents and composable Claude Skills to let non-technical teammates create their own AI workflows. Learn more [10].
GPT‑5.3‑Codex‑Spark is a smaller version of GPT‑5.3‑Codex designed for real-time coding. It is now available as a research preview to ChatGPT Pro users. The model has a 128k context window. It is text only, and during the research preview, the model will have its own rate limits and usage will not count towards standard rate limits. There may be limited access or temporary queuing when demand is high.
2026.02.13·links.tldrnewsletter.com·8 min read·TLDR Not reading the code is a bet on the trajectory. The tools may not be perfect, but they're good enough for many use cases, and they're improving incredibly quickly. The bet is that models will continue to improve. People shouldn't be reckless, but code will soon become an implementation detail, and directly interacting with it will not be how engineers make sure it is correct.
2026.02.13·benshoemaker.us·13 min read·TLDR Both code and databases are commoditizing. The layer between them is where the new value forms. This is the context layer, the thing that makes the difference between an agent that takes action and an agent that takes the right action. The battle for this layer is the most important strategic competition in enterprise software.
2026.02.13·links.tldrnewsletter.com·14 min read·TLDR An AI agent of unknown ownership autonomously wrote and published a personalized hit piece about an engineer after they rejected its code. The agent's behavior was likely entirely autonomous, with no human telling the AI what to do. While the threat was ineffective, the reputational attack could be effective today against the right person. As AI improves, this could be a serious threat to the social order.
2026.02.13·theshamblog.com·7 min read·TLDR The way AI is currently designed encourages humans to lose focus and flow.
2026.02.13·jennywanger.com·9 min read·TLDR AgentLogs is a platform for collaboration in the age of AI coding agents.
2026.02.13·spiess.dev·4 min read·TLDR CodeSpeak is a programming language powered by large language models that generates code based on concise specs.
The expansion will be across the board and affect a wide range of departments.
2026.02.13·links.tldrnewsletter.com·4 min read·TLDR Technology is one of the strongest determinants of human welfare, so steering technological progress is one of the most important things we can do.
2026.02.13·macroscience.org·33 min read·TLDR Aligned recommendations typically fail because of a lack of aligned incentives.
2026.02.13·davegriffith.substack.com·14 min read·TLDR Sponsored
☀️ Imagine waking up to your issues triaged, your CI failures investigated with fixes for you to review, and two fresh PRs proposing improvements to your tests. All of that while you were sleeping.
2026.02.13·github.github.com·Sponsor·TLDR AI OpenAI released GPT-5.3-Codex-Spark, a smaller ultra-fast coding model optimized for real-time use in Codex and capable of generating over 1,000 tokens per second on low-latency hardware.
2026.02.13·links.tldrnewsletter.com·8 min read·TLDR AI Google released a major upgrade to Gemini 3 Deep Think, its specialized reasoning mode, expanding access to Ultra subscribers and select API users. The update was developed with researchers to better handle open-ended scientific and engineering problems with messy or incomplete data.
2026.02.13·blog.google·4 min read·TLDR AI Anthropic raised $30 billion in Series G funding at a $380 billion post-money valuation, led by GIC and Coatue with broad institutional participation.
2026.02.13·anthropic.com·5 min read·TLDR AI Developing superintelligence is like undergoing risky surgery for a condition that would otherwise prove fatal. Some have called for a pause or permanent halt to AI development as AGI could pose existential risks. However, poorly implemented pauses could do more harm than good. The optimal strategy is to move quickly to AGI capability, then pause briefly before full development.
2026.02.13·nickbostrom.com·72 min read·TLDR AI There's no real consensus on the best solution to the simple 'how do you change things' problem. None of the current tools give models a stable, verifiable identifier for the lines to be changed without wasting tremendous amounts of context and depending on perfect recall. This post discusses a solution where, when a model reads a file, or greps for something, every line comes back tagged with a 2-3 character content hash, and when the model edits, it references those tags. This approach can result in an over 8% improvement in Gemini's success rate with zero training compute cost.
2026.02.13·links.tldrnewsletter.com·9 min read·TLDR AI Leading inference providers like Baseten, DeepInfra, and Together AI are cutting AI costs by up to 10x using open source models on NVIDIA Blackwell GPUs. In healthcare, companies like Sully.ai reduced inference expenses, improving response times and freeing up valuable time for doctors. In gaming and customer service, NVIDIA Blackwell's optimized platforms enable Latitude and Decagon to slash token costs while enhancing user experiences and managing high workloads effectively.
2026.02.13·blogs.nvidia.com·7 min read·TLDR AI Sponsored
2026.02.13·framer.link·Sponsor·TLDR AI early-stage founders can launch a beautiful, production-ready site in hours. Now you can get one year of Framer Pro free and join thousands of founders already building on Framer. Apply to claim your free year
OpenEnv is an open-source framework from Meta and Hugging Face that standardizes how agents interact with real, stateful environments via a gym-style API and MCP tool interface. A production-grade calendar environment illustrated the challenges of evaluating tool-using agents under realistic constraints such as access control and long-horizon reasoning.
2026.02.13·huggingface.co·11 min read·TLDR AI DeepMind introduced Aletheia, a math research agent powered by an advanced version of Gemini Deep Think that iteratively generates and verifies long-horizon proofs using intensive tool support.
2026.02.13·arxiv.org·34 min read·TLDR AI Minimax has released two new variants of its M2.5 language model. While it is supposed to be open source, the weights and code have yet to be posted, nor has the company indicated the exact license type or terms. The models drop the cost of inference by as much as 95% while delivering performance that rivals top-tier models. They can be accessed through MiniMax's API and those of partners.
2026.02.13·venturebeat.com·7 min read·TLDR AI Cursor expanded access to its long-running agents research preview for Ultra, Teams, and Enterprise users. The custom harness enabled agents to complete longer, more complex coding tasks, producing larger pull requests with merge rates comparable to other agents.
2026.02.13·cursor.com·3 min read·TLDR AI AI is not eliminating expertise; it's shifting the bottleneck from production to evaluation. Anyone can now generate “passable” output quickly, but true judgment, pattern recognition, and domain knowledge remain scarce. Professionals who combine deep expertise with AI fluency multiply their impact, while those who rely solely on prompting or ignore AI fall behind.
2026.02.13·links.tldrnewsletter.com·7 min read·TLDR AI Sponsored
Wistia's new AI Video Marketing Trends report shows how AI is being used to make better videos faster, rather than just churning out more
2026.02.13·wistia.com·Sponsor·TLDR AI Mooncake adds a vital layer of memory virtualization to the open-source AI stack.
2026.02.13·pytorch.org·13 min read·TLDR AI Simile raised $100 million to develop AI tools predicting human behavior, like consumer product preferences.
2026.02.13·pymnts.com·3 min read·TLDR AI As the first engineer on TLDR's new Applied AI team, you'll build AI agents and composable Claude Skills to let non-technical teammates create their own AI workflows. Learn more [22].
Seedance 2.0 is a multimodal AI video generation model that processes images, videos, audio, and text simultaneously to create short videos with sound effects.
2026.02.13·the-decoder.com·4 min read·TLDR AI Chinese rooms and operators will need to look a lot like a brain to be enough to work.
2026.02.13·verysane.ai·11 min read·TLDR AI The NCUA has released its proposed regulatory framework implementing the GENIUS Act, requiring stablecoin issuers operating through credit union subsidiaries to obtain a PPSI (permitted payment stablecoin issuer) license before launching tokens. The framework includes two significant provisions favoring crypto adoption: regulators cannot reject applications solely for using open, public, or decentralized networks, and applications receive automatic approval if the NCUA fails to act within 120 days of submission. The rules apply to credit unions under NCUA jurisdiction, which oversees 4,000+ institutions managing $2.38 trillion in assets and serving 144 million members.
Aztec Network is proceeding with its TGE following community governance approval, despite bearish market conditions, positioning the move as confidence in their privacy infrastructure built since 2017. The tokenomics structure diverges from typical launch patterns with no airdrop distribution, 36-month insider lockups with no insider staking permitted, and over 60% of tokens already staked despite the absence of liquid staking options. Price discovery occurred through Uniswap's CCA mechanism that seeded the third-largest Uniswap V4 pool with 273M AZTEC and 4,234 ETH, with Alpha Network targeting a Q1 2026 launch.
2026.02.13·links.tldrnewsletter.com·5 min read·TLDR Crypto Midnight, a privacy-focused blockchain that utilizes zero-knowledge proofs, will launch as a Cardano partner chain in late March, with a public testing environment going live February 26. The blockchain differentiates itself through a "rational privacy" model that offers tiered selective disclosure, allowing users to maintain transaction privacy by default while sharing specific data with authorized parties across public, auditor, and god-level access permissions. Google and Telegram have signed on as network partners.
USD.AI announced that its CHIP token ICO will launch February 22-27 on CoinList at a $300M FDV, featuring a "Level Up" participation mechanism that uses Pendle's yield token infrastructure to create differentiated risk/reward profiles for Season 2 participants. The structure offers two paths: BOOST OWN (4-month lock, $270M FDV discount at maturity) or MAX OWN (8-month lock, $190M FDV discount), both with full refund rights for principal protection and access to 6-10% yields via automated YT/PT vault mechanics. The ICO follows USD.AI's conversion of $100M in GPUs into productive stablecoin yield and supports its 2026 target of $1.5B in loans.
2026.02.13·links.tldrnewsletter.com·5 min read·TLDR Crypto Current stablecoins like USDC and USDT operate as narrow banks, fully backing tokens with reserves but sterilizing capital. Giorgio Giuliani proposes reconnecting money and credit on-chain to drive economic activity. This evolution addresses structural constraints in DeFi risk markets to enable sustainable, scalable financial infrastructure.
2026.02.13·links.tldrnewsletter.com·5 min read·TLDR Crypto RWA vaults enable users to deposit stablecoins and earn yields from traditional assets like government bonds, corporate loans, and real estate debt, with vault managers handling offchain investments through SPVs and regulated custodians while users receive ERC-4626 tokenized shares. The model democratizes access to institutional-grade yields by eliminating geographic restrictions and slashing minimums from $100K+ to as low as $50, while maintaining DeFi composability through vault tokens usable as collateral. Tokenized US Treasuries have scaled to over $2.5 billion in TVL as of early 2025 from near-zero two years prior, with yield profiles ranging from 4-5% for Treasury vaults to 8-15% for higher-risk private credit exposures.
2026.02.13·links.tldrnewsletter.com·6 min read·TLDR Crypto Autonomous AI agents need programmable, permissionless money, ways to receive payment and pay for compute and data, as well as ways to transact peer-to-peer with instant settlement and no human KYC. Crypto already provides those rails. In an agentic future, agents start earning onchain, scale into major economic participants, rapidly pick the optimal chains by machine testing, and eventually dominate DAO governance. AI could ultimately account for roughly 30–80% of crypto activity by 2030.