RETHINKING FOUNDING ENGINEER EQUITY
Founding engineers do similar work to founders and take on a similar risk (albeit a bit less), but their compensation is a lot less than other founders.
Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
Founding engineers do similar work to founders and take on a similar risk (albeit a bit less), but their compensation is a lot less than other founders.
Reasoning models show how large language models are not a dead end.
The layoffs will affect employees across multiple countries - employees will receive their notifications between February 11 and February 18.
Many of the elements that once differentiated SaaS products will lose their importance in an agent-driven future.
The war between 'GPT wrappers' will cause traditional defensibility strategies, particularly sustained advantages in distribution and network effects, to return to the forefront.
ebook to discover the best ways to make async collaboration a
Sponsored
In 2025, teams are rarely in the same physical space. Even with return-to-office mandates, work is distributed between physical locations (or ‘outsourced' to AI agents). Immediate, in-person interactions are the exception rather than the norm.
OpenAI highlights the rapid adoption of ChatGPT and the broader impact of AI, comparing it to past technological breakthroughs. Its Super Bowl ad positions AI as the next major tool in human progress, emphasizing its role in augmenting creativity rather than replacing human effort.
French President Macron announced a massive AI investment to boost innovation and international cooperation. He emphasized AI's role as an assistant rather than a job replacement, with particular focus on applications in healthcare.
SSI nearly quadruples its valuation from the previous round of funding.
This paper introduces a novel method for improving neural network training in regression tasks that involves framing them as classification problems using a learned target encoder-decoder pair. The approach outperforms traditional regression methods by leveraging the benefits of distributional target representation and smooth interpolation between different objectives.
3.5B model trained to do latent reasoning which also shows dramatic improvement on math problems. The model itself is a recurrent architecture.
While maximizing spectral gap has been used to address over-squashing in GNNs, this paper shows that minimizing it can also improve generalization. The authors propose three new rewiring strategies to enhance label-community alignment, homophily, and computational efficiency.
QuEST allows stable training with 1-bit weights and activations. QuEST achieves this by improving two key aspects of QAT methods: accurate and fast quantization of the (continuous) distributions of weights and activations via Hadamard normalization and MSE-optimal fitting, and a new trust gradient estimator based on the idea of explicitly minimizing the error between the noisy gradient computed over quantized states and the "true" (but unknown) full-precision gradient.
Agentic workflow for generating diagrams with language models.
This work introduces a new conversational framework for programming assistance that integrates coding history, current code, and user instructions.
Wiz researchers discovered unencrypted sensitive data in DeepSeek's open-source AI model, exposing major security vulnerabilities. Despite difficulties in contacting DeepSeek, the databases were secured within an hour after the issue was reported. Wiz highlighted that such security oversights indicate the service is not mature enough to handle sensitive data.
Jiayi Pan and his team at UC Berkeley recreated DeepSeek's R1-Zero model for just $30, showcasing a shift towards cost-efficient AI research. Their open-source model, "TinyZero," highlights potential challenges to big tech's expensive AI infrastructures. The development raises questions about the necessity of large-scale investments in AI.
OpenAI tested ChatGPT's persuasiveness against Reddit's ChangeMyView forum and found that its current AI models rank as more persuasive than humans in 82% of comparisons but still fall short of "superhuman" levels. The company warns of potential risks if AI achieves near-zero-cost human-level persuasive writing, as this could aid biased journalism and scams. OpenAI is implementing measures to monitor and mitigate AI-based persuasion, particularly regarding political influence and large-scale manipulation.
Sponsored
Join Elastic and Unstructured to learn the best practices for scaling the data layer that fuels your vector stores, knowledge graphs, and training pipelines. Get your GenAI apps thriving on clean canonical JSON with less stress and more efficiency for your team. Register to
Anthropic requires applicants not to use AI assistants for writing job applications to evaluate personal interest and unaided communication skills.
Meta's new Frontier AI Framework outlines conditions under which it would not release certain AI systems deemed "high-risk" or "critical-risk."
Chinese AI startup DeepSeek demonstrated its ability to replicate OpenAI's chatbots at a significantly lower cost, reigniting the open-source debate in the AI industry.
Memecoins have yet again taken center stage, with a BNB chain tutorial token (TST) reaching a $300 million market cap and getting a Binance listing after it was used in a video and mentioned by founder Chengpeng Zhao. Barstool Sports' David Portnoy promoted JAILSTOOL, which peaked at $200 million and has since fallen. Most surprisingly, the Central African Republic's president announced a national memecoin (CAR) that surpassed a $500 million market cap, though questions about its authenticity remain.
A historic liquidation event in the crypto market wiped out $14 billion in open interest within 24 hours, with ETH contracts experiencing the most volatility. Ethena successfully navigated the turmoil, capturing profits from discounted perpetual futures contracts and proving the resilience of USDe, which remained stable throughout the sell-off.
Khalani Arcadia has introduced an intent-driven protocol suite that makes blockchain-adjacent agent networks into collaborative solver networks through its Arcadia Intents Protocol (AIP). It allows for coordination between offchain agents while maintaining onchain settlement guarantees. The architecture includes “SolverFi,” a novel mechanism for handling complex strategies and multi-chain operations through specialized solver networks. Its vision is that DeFi protocols interact through intent-based coordination rather than direct integration.
Pendle experienced massive growth in 2024, with TVL increasing 20x to $4.4B and daily trading volume surging 100x to nearly $100M, establishing itself as a dominant player in DeFi yield trading. The protocol is now expanding through three strategic initiatives: enhancing V2 with permissionless listings and dynamic fees, launching Citadels to target non-EVM ecosystems and institutional markets, and introducing Boros as a new product to capture the $150B perpetual futures funding rate market. Pendle aims to become a comprehensive yield layer for both crypto-native and traditional finance users.
A new onchain analysis of Satoshi Nakamoto's wallet reveals potential connections to CaVirtEx, a Canadian exchange later acquired by Kraken. The research estimates Satoshi controls 1.096M BTC (worth over $100B), with evidence of last onchain activity in 2014, including 24 documented outbound transactions and 200 BTC sent to the Bitcoin faucet. Satoshi's connections to a centralized exchange were previously unknown and suggest that Kraken may possess relevant KYC information from its acquisition.
The article explains how to launch an AI Agent on Virtuals, discussing tokenomics, bonding curve mechanics, and the importance of using at least 40% of the token supply to "align long-term incentives". Teams must balance financial investment with operational sustainability to make sure they have enough runway for development, marketing, and liquidity while graduating from the bonding curve to a Uniswap pool.
Unvested tokens should never be stakeable since this practice undermines trust and needs to be eliminated for pre-launch projects. There should be a move away from large unlock cliffs that create market panic and instead implement smooth, linear vesting after year one. These changes would make token markets more predictable, stable, and less prone to FUD.
Memecoins once functioned like high-RTP casino games, giving casual players a real shot at big wins. However, with insiders now dominating launches and draining the odds, the game has turned predatory, stripping away the thrill and fairness that made it fun in the first place.
Many previously popular blockchain networks, like Harmony (-99.9% TVL decline), Aurora (-98.9%), Moonriver (-99.3%), and others, have seen large decreases in their total value locked (TVL), while even major chains like Polygon have suffered a 92% drop from their peak.
OpenSea seems to have taken down its token claim site, potentially because of backlash over KYC requirements in certain jurisdictions.
Of the existing cryptocurrencies without an ETF, Litecoin currently has the highest chance of approval given it's not classified as a security like SOL and XRP are.
BlackRock now owns 5% of Strategy (formerly MicroStrategy), up 0.91% since September 2024.
The Solana Foundation has launched a new coworking space with weekly events in New York City.
Opera Air is a new web browser that prioritizes user well-being through a minimalist design and integrated mindfulness features like guided breathing, meditation, and binaural beats. It includes tools such as “Take a Break” for relaxation exercises and “Boosts” for enhanced focus or creativity, seamlessly blending these into the browsing experience without disruption.
GitHub has announced new Copilot features, including Vision, which generates code from images, and next edit suggestions for more accurate code predictions. Copilot Edits allows multi-file changes with an “agent mode” for automated file selection. GitHub has also teased Project Padawan, an AI agent that handles complete software tasks under developer guidance.
OmniHuman-1 by Bytedance is an AI model that transforms images and audio into lifelike human videos. It captures natural motion, expressions, and speech synchronization with impressive realism. Trained on 19,000 hours of video, OmniHuman-1 adapts to various formats. While not publicly available, its potential raises ethical and regulatory concerns.
Artiom Dashinsky analyzes the best countries and U.S. cities for product designers to move to, considering salaries, taxes, job availability, cost of living, and quality of life. Using precise HR data from Pave, job listings from TrueUp, and cost-of-living metrics from Numbeo, he highlights how salaries compare after taxes and how job opportunities align with talent density to help designers make informed relocation decisions.
Design is shifting from rigid control to adaptive experiences. Designers' role must evolve as tastemakers, guiding AI-generated content. The focus moves from pixel perfection to dynamic, self-evolving systems. This transformation demands balancing automation with human creativity to create meaningful, personalized digital interactions at scale.
Google improved its ad quality systems to ensure that search ads lead to relevant, easily navigable pages. It introduced a new prediction model that reduced frustrating experiences by identifying ads that misdirect users. This update enhances user satisfaction and encourages advertisers to improve landing page navigation, creating a more effective and user-friendly Search experience.
This lightweight and customizable TypeScript library renders audio waveforms on HTML canvas. Create beautiful, interactive audio visualizations with ease.
Render screenshots in one simple API call instead of managing browser clusters and handling all the corner cases.
Bielke&Yang designed the Matre brand identity to honor Vatne Møbler's legacy by reissuing iconic Norwegian furniture for a modern audience. The visual identity features a bold typographic approach that uses Pressio and FKT Echo to create a confident, standout brand.
GF Smith's 2025 rebrand by TEMPLO moves beyond its monochrome heritage, embracing bold colors, a custom typeface, and a dynamic new identity. The transformation reflects its commitment to creativity and community while positioning the company for the future.
Even when identical to human advice, medical advice generated by AI was rated less reliable and empathetic. However, AI-made ads received higher ratings. Human-AI teams often excel in creative tasks. Trust in AI depends on user experience, domain, and education. There is greater resistance towards AI use in sensitive areas like healthcare and employment.
Finnish designer Yrjö Kukkapuro, known for his ergonomic and postmodern furniture, has passed away at 91.