Full-text search across 8183 articles. Combine with topic and date filters; results sorted by relevance.
Anthropic's Claude Code now includes an in-app browser that lets developers navigate live web pages, take screenshots, annotate elements, and bring visual context into coding workflows without requiring external APIs or …
The context sandwich is a three-part prompting framework that wraps a task with background context and output expectations, improving AI responses from models like ChatGPT and Claude. The method, explained by an unnamed …
Zack Proser built ctx, a personal context engine that lives outside any single AI tool or vendor, to solve the problem of having to re-explain his life and work to every new agent. ctx is a single Postgres table behind a…
LoRA fine-tuning allows enterprises to customize AI models on proprietary data at a fraction of the cost of full retraining, reducing trainable parameters by up to 10,000x. The technique, introduced by Microsoft research…
Local open-weight models and PII detection pipelines can process sensitive documents without sending data to the cloud, eliminating risks such as data retention, breach surface, and regulatory liability under HIPAA, GDPR…
AI infrastructure is shifting from token-based measurement to verifiable work units that bundle acceptance conditions, execution evidence, total cost, and accountability boundaries, as signaled by recent product releases…
Together AI and Y Combinator have partnered to launch the first dedicated GPU cluster for Y Combinator's portfolio of AI-native startups, providing flexible, short-term access to compute at long-term rates. The cluster, …
Usage-based billing in AI products uses tokens, credits, or outcomes as the unit of consumption, each shifting risk differently between customer and vendor. Tokens, the cost unit of AI, are counted per request and priced…
Thinking Machines, led by Mira Murati, released Inkling, a 975B open-weights model that companies can fine-tune via its Tinker platform without building GPU infrastructure, enabling startups to own specialized AI models.…
Anthropic's Claude Fable 5 model, live for 40 days, leads competitors on SWE-Bench Pro by 11 points with an 80.3% score and costs $10/$50 per million input/output tokens, but its refusal system returns HTTP 200 with stop…
Reinforcement learning (RL) is the primary driver of recent advances in coding and mathematics for large language models (LLMs), according to an analysis of the economics of model improvement after human-curated data has…
Georgia Tech's Systems Software and Security Lab has confirmed 74 CVEs directly attributed to AI-generated code in 2026, with the actual number estimated at 400 to 700 cases in the open-source ecosystem alone. Apiiro's a…
A 4-step roadmap published by a tech career guide outlines how employees can become in-house AI consultants by auditing their job for automation opportunities, building solutions, demonstrating ROI, and formalizing the r…
SkillsBench, an arXiv benchmark published in February 2026, found that curated skills boosted average task pass rates from 33.9% to 50.5% across 87 tasks and 18 model-harness configurations, with healthcare jumping from …
A developer built a DeFi trading agent using the Model Context Protocol (MCP) that can browse pair markets, read positions, and execute leveraged pair trades on Hyperliquid via Pear Protocol. The open-source mcp-pear ser…
VS Code 1.129, released July 15 by Microsoft, introduces the Agent Host, which moves AI coding agents into a dedicated background process, preventing agent failures from freezing the editor and enabling persistent, multi…
A developer built an offline-first sync engine in Rust, using event sourcing to reconcile data between mobile devices and a backend. The engine, shared across iOS and Android via UniFFI, stores events rather than current…
Claude Code can register an MCP server with a single CLI command, and for many teams the right number of MCP servers is zero—a shell command in a rules file does the same job with less context overhead. The minimal setup…
A full-stack engineer has built VulneraMCP, an AI-augmented security testing platform that uses the open-source ZAP scanning engine with a machine learning layer for bug bounty hunting. The system integrates ZAP's REST A…
A developer built ToolHail, a discovery-and-execution layer for AI agents that lets them find and use MCP servers without human intervention. ToolHail is a remote MCP server providing two tools: find_mcp_server for disco…