Auto-Tune Your LLM Judge
A new methodology for tuning LLM judges focuses on reducing variance rather than just accuracy, as evaluators can flip decisions up to 15 points across runs. The approach, outlined by an unnamed author, involves measurin…
Full-text search across 68380 articles. Combine with topic and date filters; results sorted by relevance.
A new methodology for tuning LLM judges focuses on reducing variance rather than just accuracy, as evaluators can flip decisions up to 15 points across runs. The approach, outlined by an unnamed author, involves measurin…
Spotify has launched Xirp, a vendor-neutral agentic development environment that connects AI coding tools to a company's services, ownership, and architectural decisions, aiming to solve the retrieval problem of institut…
Google is preparing to release Gemini 3.7 Flash, its next lightweight AI model, as code references have surfaced in the company's Python SDK on GitHub, just weeks after the July 21, 2026 launch of Gemini 3.6 Flash. The 3…
AI Agent engineering roles now pay 30–50% more than frontend positions with comparable experience, according to early 2025 China market data, as junior and mid-level frontend work is automated by tools like Copilot, Curs…
Neo4j's Graph ML and GenAI Research team introduced the Meta Knowledge Graph (MKG), a graph-structured context layer that supplies AI agents with operational metadata—semantic descriptions of enterprise data and memory o…
A senior software engineer and tech lead has published a guide to creating Architecture Decision Records (ADRs) specifically designed for AI agents, drawing on concepts from Dan Shapiro and Nate B. Jones. The ADRs includ…
AWS released Dogwood on August 6, an open-source policy language extending Cedar to govern sequences of AI agent tool calls within a session, available under Apache 2.0 on GitHub. Dogwood adds temporal operators such as …
OPSWAT's AI Content Inspector adds a content authenticity verdict to its MetaDefender pipeline to detect AI-generated invoice fraud, which produces convincing fake invoices that pass standard malware and PDF security sca…
Software developers are increasingly using AI agents to generate code, which they then refine by removing unnecessary parts, a process likened to sculpting. This approach, enabled by AI, allows developers to quickly prod…
Meta released Muse Glimmer on Monday, August 10, a 30-billion-parameter open-weight AI model that runs on a single consumer GPU, available for free download from Hugging Face under an Apache 2.0 license. CEO Mark Zuckerb…
A developer working on the Algebraic Architecture Theory (AAT) research project has introduced a philosophy section into the project's AGENTS.md file, outlining ethical and methodological principles for AI agents. The ph…
DeepSeek-V3, an open-weight large language model from Chinese AI company DeepSeek, has leaked and is matching or beating top-tier proprietary US models in coding and math benchmarks, according to a technical review. The …
NVIDIA released its Alpamayo self-driving models as open-source code for commercial use, claiming 'exceptional performance' for the new Alpamayo 2 Super model on the LingoQA test. The move targets Japanese automakers lik…
Developer mstdokumaci released zsort, an opinionated import organizer for the Zig programming language, inspired by Python's isort and available on GitHub. The tool uses std.zig.Ast and supports Zig versions 0.15.2 and 0…
A CEO overwhelmed by AI projects shares a system for staying focused, including defining 'done' before starting and handing off projects to team members. The system addresses the tendency to fall into AI rabbit holes, as…
Databricks reports that large enterprises can cut AI coding costs by up to 90 percent by using open models, efficient routing, and other techniques, based on its internal data and conversations with Stripe, Coinbase, Ube…
Anthropic's Opus 5 model upgrade prompt, released for Claude Code, audits and upgrades existing skills through a blind-judged Gauntlet Loop, retiring those that don't outperform the base model. The prompt, which requires…
Figma's latest earnings show the company continues to thrive, reinforcing that SaaS remains viable in the AI era, while Amplitude also reported strong quarterly results. Meta introduced a new coding-focused AI model, Met…
Netlify now supports open-weight AI models including DeepSeek, Qwen, GLM, and Kimi through partnerships with OpenRouter and OpenCode, giving every customer full model freedom via Agent Runners and AI Gateway. The company…
Red Hat launched asago, an open source project under the Apache 2.0 license, to automate translating AI governance policies into operational controls, generating safety tests and recommending guardrails. The framework ma…