Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
Meta's Muse Image model can generate images grounded by search. It can reason before it renders. The model is priced for production volumes at $0.01 per image. Real-world examples of Muse Image workflows are available in the link.
Microsoft AutoSaddler is a system that analyzes agent execution traces and automatically updates prompts, tools, and middleware to improve agent performance.
WeChat released WeMM-Embedding, a family of multimodal embedding models that maps text, images, videos, visual documents, and interleaved inputs into a unified representation space.
Barret Zoph was a co-founder at Thinking Machines Lab before defecting to OpenAI this year due to a dispute with Chief Executive Mira Murati. Zoph is now leaving OpenAI to join Google as vice-president of research. Google is currently restructuring its AI development efforts to catch up with rivals in developing AI that can create code. Zoph had worked as a research scientist at Google from 2016 through to 2022.
Anthropic has inked a roughly $45 billion cloud deal with Nscale, a UK-based AI infrastructure company. Anthropic will rent around 460 megawatts of compute capacity at an Nscale data center development in West Virginia. The facility, which is expected to come online at the end of 2027, will have Nvidia's advanced Vera Rubin chips.
Sponsored
Agent traffic grew 7,851% last year and the share getting blocked tripled in a single month, mostly with no policy behind it. Read an Agentic Survival Guide [18]for governing agents instead of blocking them written by cybersecurity expert Daniel Miessler.
Claude's internal browser is rolling out this week to Pro, Max, and Team plans in the Claude desktop app.
Grok Bot now comes with SuperGrok and Cursor plans, offering expanded access without affecting existing usage limits.
Bill Gates, once optimistic about AI, now warns it could become a major source of injustice and cause widespread unemployment.
Rather than adding experts, the model bolts on 51B parameters as a separate embedding indexed by two- and three-character fragments.
Mechanize develops virtual environments, benchmarks, and training data to help AI agents handle more complex tasks.
Shinhan Financial Group signed a strategic agreement with Visa to build stablecoin-based financial infrastructure, using Visa's platform to verify issuance, remittance, and redemption while co-designing a Korea-specific model. The partnership will pilot stablecoins in card payment settlement, AI-based payment models, and B2B/B2C payments, combining Shinhan Bank, Shinhan Card, and Jeju Bank with Visa's network. It follows Shinhan Card's April partnership with Solana Foundation and an August MOU with Solana, Etherfuse, and Orca to test a won-denominated tokenized fund, as South Korea works on its Digital Asset Basic Act.
Bitcoin held near $79,000 on August 26 as traders booked profits following a week that delivered a 23% BTC gain and a nearly 45% XRP rally. Ether and Solana each fell approximately 4% over 24 hours, with most major tokens tracking lower on the day. HYPE was the lone exception among large-cap tokens, avoiding the daily decline. The daily softness in altcoins following those weekly advances is consistent with short-term distribution at elevated price levels.
Ethereum's speed directly drives its economic value by reducing friction (idle capital, wait times, and settlement costs) for payments, swaps, and credit. There are four active workstreams: "fast blocks" (removing the hardcoded 12-second block time, starting with a proposed cut to 10 seconds in the upcoming "Hegotá" fork); "fast finality" (an EF/Consensys proposal to cut finality from ~15 minutes to a couple minutes in the following fork); the "Fast Confirmation Rule" (near-production today, cutting confirmation to tens of seconds with formally provable security, already adopted by one major bridge); and "fast interoperability" via real-time zk proving to speed up L2-to-L1 asset flows. Hegotá is possibly the last practical window to kick off block-time reductions for years.
Bitwise has launched Automated Token Portfolios (ATPs), institution-grade model portfolios built on Coinbase's tokenized equities (NVDAc, METAc, AAPLc, and GOOGLc) deployed on Base two days prior. The structure is non-custodial throughout: Bitwise defines the target allocation, users hold tokenized stocks in their own wallets, and Glider's automated rebalancing protocol maintains alignment without taking custody. The product brings TradFi model-portfolio infrastructure to on-chain self-custody, giving non-US investors access to institution-grade equity exposure without surrendering asset control. Access is restricted to non-US persons under Regulation S, confining the initial addressable market to eligible jurisdictions outside the US.
The "coffee vs. matcha" framing (stablecoins vs. tokenized deposits) misses that the two financial tools can be used together. A tokenized deposit is a direct bank liability suited for treasury and payments within a controlled institutional perimeter, while a stablecoin is a separate liability designed to travel across open networks. The more likely architecture has tokenized demand deposits forming the operational liquidity layer underneath a stablecoin's reserves. The stablecoin issuer holds a deposit claim on the bank, and minting/burning could synchronize atomically and 24/7 with reserve receipt and redemption, rather than through today's manual orchestration.
EIP-8130 is the correct native account abstraction path for EVM rollups, encoding the three primitives users need (transaction batching, gas sponsorship, and key rotation) without the complexity of EIP-8141's seven new opcodes, frame modes, and layered execution model. EIP-8130's deterministic contract addresses enable cross-chain account changes without bridges, and wallets receive a canonical interface for secp256k1, P-256, and WebAuthn/passkey authenticators. With Ethereum's Glamsterdam upgrade slipped to Q4 2026 and EIP-8141 not credibly activatable before late Q2 2027, L2s can adopt EIP-8130 now. Base is targeting EIP-8130 in its Cobalt upgrade for a 2x per-transaction cost reduction, with Optimism partnering to extend native AA across the OP Stack.
x402 hit 1-1.5 million transactions per day for the first time, with volume up ~2x versus June/July, but average transaction size has fallen to $0.023-$0.031. Since popular third-party facilitators (Coinbase CDP on Base, PayAI on Solana) charge a flat $0.001 per settlement, that fee now eats 32-43% of average transaction value. Batching, live since May, could cut that cost by 90-99% by settling thousands of transactions in one, but adoption remains near zero. There remain questions around whether tooling, counterparty risk, or accounting complexity is the blocker.
Airwallex, which moves cross-border payments across 120+ countries via local currency pools and same-day local network settlement, is integrating stablecoin support after years of deprioritizing it, citing AI agent microtransactions and US and European regulatory clarity as the primary catalysts. The company's existing rails cover most enterprise payment needs, but card processing fees make sub-dollar transactions uneconomical. Stablecoin wallets with fixed balances solve both the unit-economics problem and spend-control requirements for agent-driven API calls, compute purchases, and data lookups at scale. To build out the required settlement layer, Airwallex invested to incubate Metal, a settlement infrastructure product built for regulated financial institutions to process transactions at agent speeds, while also adding direct stablecoin support in markets where regulators permit it. The firm positions agent payments as the largest source of near-term stablecoin payment demand, with slower clearing corridors as a secondary use case for recipients able to hold stablecoins under local regulation.
Wirex One (+59%) and Plasma One (+42%) posted strong 30-day spending growth, while Bitget Wallet (-49.2%) and SafePal (-62.2%) saw sharp declines.
Wirex told users its cards aren't subject to DAC8 reporting (effective January 2026) because Wirex isn't custodian of user funds.
Japan's FSA, Ministry of Finance, Bank of Japan, and financial institutions plan to form a study group this summer to design blockchain-based real-time settlement infrastructure for government bonds and stocks, targeting a development plan by early 2027.
Grayscale's ZCSH launched as the world's first Zcash ETF, giving traditional brokerage accounts direct exposure to ZEC, a proof-of-work asset with Bitcoin-equivalent tokenomics (21M cap, halving schedule) plus optional shielded transactions.
Sponsored
They're all showing up to OpenSearchCon North America [4] to share how they run AI search, observability, and data infrastructure at (massive) scale. Come for real production patterns and two hands-on workshops: * Build an OpenTelemetry observability stack * Build an agentic AI search app Additional speakers [4] come from Airbnb, Intel, IBM and more. Join in San Jose from September 22 to 24, to build the future of open source search. Come for the talks, leave with something you can actually use on Monday.
LinkedIn built a cognitive memory agent that personalizes its hiring assistant using four layers: conversational, episodic, semantic, and procedural memory. It replaced GraphRAG with a hierarchical tree that supports cheaper incremental updates, while dedicated ingestion, consolidation, and retrieval services keep preferences fresh, traceable, access-controlled, and available within tight latency limits.
DoorDash's SafeChat handles millions of chats, calls, and images per day with a hybrid moderation pipeline. Fast internal classifiers clear obvious safe cases, LLMs score the hardest cases across severity axes, and no-code workflows plus backtesting let teams tune actions before production. The system cut verbal-abuse incidents roughly in half.
In Kafka-style streaming systems, the shared log is the ordered event backbone that brokers and consumers build on. LogDrive separates that log into an AtomicLog sequencing layer and a durability layer built from flat single-value registers plus weakTail, making striping and quorum replication reusable across S3, DynamoDB, S3 Express, or other backends.
RocksDB amplification depends heavily on workload shape, compaction capacity, and key order. Random inserts produced 4.56x write amplification while sequential keys stayed the same, and delete-heavy workloads temporarily pushed space amplification to 1.58x. Bloom filters kept L0 read latency nearly flat as file counts grew.
GitHub's secret-scanning work shows why LLM evaluation should optimize the production decision, not generic benchmark scores. Prompts, models, input construction, and eval configs were versioned like code, then tested against production-like ambiguity. That discipline cut false positives by 95% offline while preserving recall within the safety guardrail.
Complex tasks can exceed a single AI agent's organizational capacity, regardless of better context or tools. Graph Engineering addresses this by mapping tasks, specialized agents, and system states to coordinate parallel work, verification, and persistent execution.
AI creates more pipelines and dependencies without matching growth in data teams, making engineers the operational bottleneck. Companies should prevent burnout with scalable infrastructure, governance, and automated operations matched to their actual needs and capacity.
Sponsored
Missed RFS dates, Phantom Clusters, and layers of vendors compound challenges that start with a volatile supply chain. Michael Francisco, Vice President of WhiteFiber Cloud US, breaks down what's driving these dynamics from the seller's side, and how to simplify your sourcing process.
DuckLabs is joining AWS to gain the infrastructure, resources, and reach needed to scale DuckDB and the wider Duck Stack. DuckDB, DuckLake, and Quack will remain MIT-licensed open-source projects stewarded by the independent DuckDB Foundation, while AWS supports faster development and broader adoption.
Khaos is an open-source Kafka load-testing and chaos-simulation tool that was recently rewritten in Go. It generates realistic producers and consumers, exercises broker failures, hot partitions, consumer lag, rebalances, and backpressure, and gives platform teams a repeatable way to test Kafka behavior before they encounter production incidents.
PostgreSQL regex search can vary wildly by index and extension. In a 1.6M-row, 33 GB plan table, pg_trgm cut a simple LIKE scan from about 40 seconds to 1.6 seconds, pg_tre added fuzzy matching but had painful index build time, and pg_re2 delivered faster regex execution with fewer language features.
PagedAttention brings virtual-memory-style paging to LLM KV caches by replacing contiguous per-request allocation with fixed-size blocks and a block table. That design eliminates most cache fragmentation, raises utilization to roughly 96%, supports copy-on-write for shared prompts, and can serve 2–4x more users per GPU without changing attention outputs.
Enterprise application software is still growing, but aggregate market numbers hide where AI is actually changing buyer behavior. Workflow-heavy and automation-friendly categories are pulling ahead while customer service and HCM lag. For buyers, AI packaging and pricing now reveal more about value capture than broad market-growth headlines.
Analyst Gym uses daily two-minute workplace drills to sharpen analytical judgment rather than coding skills.
Neocarta builds a Neo4j semantic layer that unifies database schemas, business definitions, metrics, and query history, then exposes this context to AI agents through MCP.
Calendly has unveiled a new logo, wordmark, and color system alongside an AI email assistant and an AI meeting notetaker. The rebrand replaces the previous neon palette with soft prismatic gradients and a rippling "gloop" motif, signaling that the product is more than a scheduling link as AI reshapes productivity software.
Google has refreshed the Clock and Pixel Magnifier app icons as part of the visual design language introduced with the Pixel 11. The new Clock icon features thicker, watch-inspired hands, a blue gradient face, and a white background that aligns it with Google's broader app ecosystem, while Pixel Magnifier receives a similar treatment with a thicker gradient magnifying glass and updated Quick Settings icon. These changes are purely cosmetic, with no notable updates to the apps' functionality or user interfaces.
Apple has begun rolling out ads in Apple Maps across the US and Canada, showing sponsored locations before and after searches. While Apple says the ads are privacy-focused and don't use personal tracking, critics argue they still risk undermining Apple's privacy reputation and make its premium products feel less premium by introducing advertising into built-in apps.
AI has made finished-looking output cheap, so credibility at work now rests on the trail behind an artifact rather than the artifact itself. Contributors, managers, and agents each need a different ledger: rejected directions, decision rationale, and scoped, reviewable execution. A useful one captures itself from tools already in use and records only what explains how the work came to be.
The first version of ChatGPT memory was a simple notepad the model could read from, write to, and edit across conversations. The hard part was not the interface, but training the decision boundaries for what is worth saving and when to recall it. Most current memory implementations still over-apply what they know, which makes productive forgetting the missing capability.
Apple's rumored foldable “iPhone Ultra” is reportedly impressing early testers with its pocket-friendly design, durable hinge, and larger iPad-like internal display, potentially making it Apple's most exciting hardware launch in years. However, leaks suggest it may omit key premium features such as a telephoto camera and Face ID, raising questions about its value at a rumored $2,000 price point. For creatives, the choice may come down to whether a larger foldable screen is more important than having Apple's best camera system.
Sponsored
Making a screen is easy. Getting the UX right inside an existing product is harder. Figr learns your design system, guidelines, flows, and analytics first.(See what Figr designed for products you already know) [11] Then map states, test edge cases, write PRDs, and create screens you can move into Figma with Auto Layout, components, and tokens intact.