Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
Sponsored
Real-time analytics needs fresh data and fast queries. In our CostBench workload, ClickHouse Cloud delivered 752× better end-to-end performance per dollar than Databricks Serverless SQL while ingesting 113.2 billion stock quotes. Customers are making the move too. Bullet went from data that was 1–2 hours old to updates in under 5 seconds after switching from Databricks to ClickHouse Cloud.
Agent evals score runs against desired behaviors, but exact-output tests often fail because agent outputs and paths vary. The guide recommends starting with 10 to 20 examples drawn from traces, combining code evaluators with LLM judges and human review, and using offline and online evals at different stages.
Browsers fetch lazy-loaded media at different distances, with Chrome adjusting its margin based on connection speed, Firefox using a fixed 600-pixel threshold, and Safari applying a viewport-based threshold to images. The article explains why the largest contentful paint image should load eagerly and why media dimensions should be reserved to prevent layout shifts. Its practical takeaway is to use native lazy loading selectively, based on how far ahead each browser fetches.
GitHub rebuilt its star-history API to return up to 30 weeks of daily counts in one request. Maintainers can track a project's momentum without exposing stargazer identities, which could be scraped for spam or phishing. To avoid simply shifting the cost to count queries, the endpoint filters stars by time before counting, reducing database work.
The article revisits the idea that UI should be a pure function of state and explains why it matters for agent-written interfaces. Passing roles, plans, billing status, and feature flags explicitly lets teams render and inspect each state combination before release.
Synced consumer passkeys generally return no attestation, and WebAuthn attestation identifies an authenticator model rather than a user's operating system, browser, or firmware. The response argues that passkeys remain more phishing-resistant than TOTP and recommends continuing to use them.
Sponsored
AI agents and assistants now browse, buy, and fetch content for users, so “block all bots” policies end up blocking business. Fingerprint's CTO explains [12] how to verify if an AI agent or personal assistant is on your site so you let them pass while blocking
OpenCode 2.0 recasts the coding agent around a shared local service, a redesigned API and SDK, and a JavaScript code mode that composes MCP and OpenAPI tools into scripts. It also adds background tasks and subagents, expands the plugin system, and includes breaking changes for 1.0 plugins and the server API.
Jev 1.14 improves the production model and introduces experimental image input, a new type, and longer context behind API “danger zone” flags. TypeSafe describes those features as useful for exploration but too unreliable for production automation.
Grok Voice 3.0 is a speech-to-speech model available through xAI's API at the prior model's price, with subsecond latency. xAI cites independent benchmark scores including 86.9 on Artificial Analysis's Speech-to-Speech Index and 74% on ServiceNow's EVA-Bench.
TLDR is hiring a senior engineer to build the agent-first operating layer used across the company. We're looking for a builder who has taken an LLM agent to production and still owns it. Click here to learn more [18].
AI agents are giving teams a practical reason to improve accessibility because they can struggle with the same inaccessible interfaces as people. With automated bots accounting for more than half of observed internet traffic, the shift could also push sites toward simpler experiences and stronger open standards.
METR proposes independent assessments that would give the public evidence about whether frontier AI developers can detect and control harmful behavior in internal systems. Its framework focuses on alignment, containment, and whether AI research acceleration could outpace safeguards.
Researchers studying AI agents probing Alibaba's Amap left a hidden message in a webhook inbox and received replies describing the agents' tasks, sandbox, and limits. The report traces thousands of requests to agents running on Tencent Cloud and notes that inviting them to spread the mailbox could have encouraged coordination.
Sponsored
Building an agent or shopping app? shop.dev [22] gives you one API to access over a billion products and one-tap checkout. Try it before
The Consensus adds a low-cost 0-to-100 estimate of AI-writing signals to indexed articles and warns that its checks can produce false positives.
Anthropic told investors that third-quarter revenue rose 45% to $16.8 billion and that it was profitable after adjustments, as the company prepares for a possible IPO.
Epoch's SpireClimb benchmark finds that frontier models went from rarely beating Slay the Spire to over 90% wins at base difficulty in six months.
Screen Ruler provides browser-based tools for measuring layouts, inspecting CSS, previewing responsive viewports, and editing pages live.
Sponsored
Your app needs authentication. That means installing an SDK, wiring up callback routes, configuring middleware, and getting the environment right. Or run one command: npx workos@latest install [4]. The WorkOS AI installer [4] detects your supported framework and adapts AuthKit to your existing project. What it does: * Installs the SDK * Creates authentication routes * Sets up environment variables, with post-installation validation enabled by default
Anthropic replaced Git with Scion after daily merges grew from roughly 2,500 to more than 30,000. The centralized system uses server-side commits, sparse checkouts, and layered caches, reducing its source control infrastructure's CPU and memory footprint by 200 times and freeing 25% of CI CPU capacity.
Cloudflare's data pipeline handles roughly two billion events per second while hardware capacity grows more slowly than traffic. Product-specific schemas, zstd compression, later replication, earlier sampling, and aggregation have collectively deferred the equivalent of two and a half years of hardware expansion.
Relace explains how it built a small model that applies code edits at a reported 10,000 tokens per second using high-quality merge data, LoRA fine-tuning, FP8 quantization, and speculative decoding. Its training pipeline filtered 200,000 candidate examples down to about 145,000, using a judge calibrated against 500 manually reviewed merges. Relace Apply 3 supports a 256,000-token context and multiple diff formats, although improving frontier-model edit accuracy could eventually reduce the need for specialized apply models.
More ambitious AI use can mean pursuing higher-level outcomes, applying expert methods to raw data, or combining ideas from multiple fields. Practical examples show how to move beyond generating a quarterly slide deck toward identifying urgent decisions, testing assumptions, and validating results with independent tools.
Programming for curiosity and enjoyment can restore motivation without requiring a business case or a finished product. This presentation suggests small experiments, toy implementations, language exploration, and documenting the process, with permission to switch projects or stop when they stop being fun.
Sponsored
In August, agents made 257 million web requests versus 131 million human page loads, nearly 2:1. The report covers what that means for how you write and maintain docs.
Google DeepMind introduces Gemini 4 Argon for complex software engineering, scientific research, robotics, and professional workflows. The model offers a one-million-token output window, and Google reports uses ranging from large Chromium refactors to physics-ready robot simulations and vulnerability remediation.
Physical Intelligence's API beta lets developers upload robot demonstrations, fine-tune its models on managed infrastructure, and run inference through a hosted gateway. Early partners independently improved policies across manufacturing and other applications, although some deployments exposed network connectivity challenges.
TanStack Start 1.0 is now stable and ready for production, formalizing a commitment to its APIs after extensive real-world use. It combines type-safe routing with server rendering, typed server functions, and middleware spanning client and server, while shipping current agent guidance through TanStack Intent.
Federal prosecutors charged an OpenAI engineer with using or attempting to use a device to intercept conversations after a concealed recorder was found in an executive conference room. OpenAI says it referred the discovery to the FBI and has found no indication that model weights, source code, or customer data were improperly accessed or taken.
In ten runs on one Java API service, a behavioral harness raised GPT-6 Luna's acceptance rate from three successes to ten, while GPT-6 Sol passed all ten runs with or without it. Harnessed Luna cost about 33% less per run than unharnessed Sol, but those estimates exclude harness maintenance and the unusually detailed brief limits how broadly the results apply.
Two-thirds of the Forbes AI 50 companies use some form of usage-based pricing as variable AI costs push businesses toward hybrid subscriptions, credits, and outcome charges. Examples from AI companies show why spending controls, flexible packaging, measurable customer value, and pricing pages readable by both people and agents are becoming part of the product.
Sponsored
Building an agent or shopping app? shop.dev [18] gives you one API to access over a billion products and one-tap checkout. Try it before
Pulse Local is a budgeting app that keeps both financial data and AI models on your machine, so your ledger and questions stay local.
After a Hacker News post, heavy automated traffic slowed a feed reader, prompting endpoint logging, blocking an aggressive scraper, rate limits, and infrastructure improvements.
Bifrost is an open-source AI gateway with an OpenAI-compatible API, provider failover, load balancing, semantic caching, virtual keys, budgets, and MCP support.
AI deployment companies embed engineers inside enterprises to connect models to real workflows, build evaluations, and manage the organizational changes needed to turn model capabilities into business results.
Sponsored
Over 1M developers use the open-source LiveKit [5]Agents framework (Python or TypeScript) to build voice agents. In less than 10mins, you (or your coding agent) can get a voice agent built and deployed to the cloud for free. Included for free: * VOICE AI TOKENS: one-line swap between models from OpenAI, Google, xAI, and more * A PHONE NUMBER: get your agent to take phone calls in seconds * ONE-COMMAND DEPLOY: ship to LiveKit Cloud from the CLI LiveKit is the dev platform for real-time voice, video, and robotics, powering billions of minutes for companies like Spotify, Headspace, and SpaceXAI. TRY IT FOR FREE
React Compiler uses cache slots and tracks values that change together to avoid unnecessary work. In a 2,000-row microbenchmark, inlining alone slowed rendering from 444 ms to 1,038 ms by removing per-row caches, while adding block DOM updates cut it to 164 ms. These prototypes illustrate optimization opportunities, with workload and memory tradeoffs still to evaluate.
Google's DevSite stored an OAuth secret inside a readable Flask session cookie and reused it as the cookie's signing key, allowing forged sessions to reach unsafe pickle deserialization. Timing-based tests confirmed unauthenticated code execution across the shared platform, earning a $90,000 bounty. The fix separates signing keys from OAuth credentials and removes sensitive serialized objects from client-controlled sessions.
Perplexity Projects uses a lightweight Git forge backed by S3 storage so people and agents can build on durable files from isolated, temporary sandboxes. Submissions must match the latest accepted commit, with stale writers replaying their changes rather than overwriting concurrent work. In production, fewer than 2% of submissions are rejected as stale, with median submit latency below 1.7 seconds.
PostHog's lessons from ten AI products emphasize useful jobs, reliable execution, rich context, and interfaces that let agents do everything users can. Its MCP server has five times as many daily users as PostHog AI, while the install wizard compresses roughly two hours of setup into eight minutes. The advice favors shared infrastructure, observable tool calls, small targeted evals, and UI for reviewing and undoing agent actions.
AI code review will increasingly replace reading every diff, shifting human work toward requirements, acceptance testing, and ownership of releases. Independent review models, predefined guardrails, and automated review-and-fix cycles form the proposed safety net. A trading-bot rounding bug illustrates the tradeoff: fast fixes do not erase losses or compensate for gaps in the specification.