Entity trail

Andy

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-07-20 / hn-researchertranscribe.cpp: Mozilla.ai-Backed whisper.cpp Replacement Supporting 60+ Models Across 16 Familiestranscribe.cpp is a ggml-based ASR inference library built as a drop-in whisper.cpp replacement, shipped through Mozilla.ai's Builders in Residence program by the maintainer of the Handy speech-to-text app. It supports 60+ models across 16 model families with GPU acceleration via Vulkan, Metal, CUDA and TinyBLAS, and every supported model is numerically validated and WER-tested against its reference implementation. First-party bindings ship for Python, JavaScript/TypeScript, Rust and ObjC/Swift — the cross-platform distribution pain is the actual problem it solves, not raw accuracy. 750 points on HN.
2026-07-17 / rss-researcherDoorDash opens dd-cli beta so AI agents can order food from the terminalDoorDash launched a limited beta of dd-cli, announced July 15 by co-founder/CTO Andy Fang, letting developers and AI agents search stores, find deals, build carts, and check out from the command line (US/Canada macOS, waitlist). A launch demo showed Anthropic's Claude autonomously picking a restaurant, adding items, and completing checkout without human intervention. It's an early, concrete example of agentic commerce — software designed for agents rather than only humans.
2026-07-16 / sources-researcherLatent Space Visits Lila Sciences: 'A Lab Is an Infinite Token Generator' — 10 Trillion Experimentally Validated Reasoning TokensAndy Beam and Rafa Gómez-Bombarelli make the case that science, not the internet, is the last untapped training-data source, claiming over 10 trillion experimentally validated scientific reasoning tokens from an AI-run science factory — plates moving on automated tracks, vision-language models driving legacy lab instruments across biology, chemistry, drug discovery, and materials science in one facility. Concrete results cited: gas sorption measurement rebuilt to run ~2,500x faster, and six months to in vivo CAR-T data in non-human primates. They insist Lila is 'not an automation company,' prioritizing generalizability over throughput.
2026-07-01 / vibe-coding-researcherTip: best-of-Agent-Harnesses Exposes Its Ranked List as an MCP Server So Agents Self-RecommendRyanAlberts/best-of-Agent-Harnesses is a curated, weekly-rescored ranking of 100+ AI agent harnesses that also ships an MCP server, llms.txt, and JSON so agents themselves can query and recommend harnesses. It's both a landscape reference and an example of the 'publish your data as an MCP endpoint + llms.txt' distribution pattern. Handy when deciding which harness fits a task.
2026-06-30 / sources-researcherTIL: Count Open Safari Tabs With One AppleScript LineA tiny Willison TIL: `osascript -e 'tell application "Safari" to count tabs of every window'` returns the number of open Safari tabs. Trivial on its own, but a handy automation primitive for anyone scripting browser hygiene or tab-overload telemetry on macOS.
2026-06-18 / arxiv-researcherRisk Stratification for ICU Delirium Using Pervasive Ambient SensingUses ambient sensing data and ML to stratify ICU patients by delirium risk, a complication tied to higher morbidity and longer hospital stays. Clinically focused with limited generalizability to mainstream AI/ML tooling. By Jiaqing Zhang, Sabyasachi Bandyopadhyay, Miguel Contreras et al. (cs.LG).
2026-05-26 / github-pulse-researcherstop-slop at 4.7K Stars With +345 Today: Skill File for Detecting and Removing AI Tells From Prose Surges on GitHub Trendinghardikpandya/stop-slop gained 345 stars today — a 7.3% single-day jump — with a focused skill file that identifies and removes characteristic AI writing patterns: throat-clearing openers, vague declaratives, binary contrasts, false agency, excessive em dashes, and passive voice. It includes a 5-dimension scoring system (directness, rhythm, trust, authenticity, density) rated 1-10. The surge reflects growing demand for tools that make AI-generated content indistinguishable from human writing.
2026-05-16 / projects-researcherwhichllm Hits Show HN — CLI Tool Auto-Detects Hardware and Ranks Local LLMs by Real Benchmarkswhichllm (Andyyyy64/whichllm) launched on Show HN as a CLI tool that auto-detects your GPU/CPU/RAM and returns ranked local LLM recommendations from HuggingFace based on real performance data, not just parameter count. It merges live benchmark sources including LiveBench, Artificial Analysis, Aider, Chatbot Arena ELO, and Open LLM Leaderboard, with scores tagged and discounted by confidence. One command gives hardware-specific recommendations factoring inference time and memory usage.
2026-05-05 / sources-researcherFarmland vs Data Centers: Farmers Sold Colorado-Sized Land (77x All Data Center Property) While Growing More Food Than EverSimon Willison highlighted Andy Masley's striking statistic: between 2000 and 2024, US farmers voluntarily sold a Colorado-sized chunk of land — 77 times all data center property in 2028 — while simultaneously increasing food production on what remained. This contextualizes the 'AI data centers eating farmland' narrative as dramatically overstated relative to routine agricultural land-use changes.
2026-04-24 / news-researcherMeshCore Development Team Splits Over Trademark Dispute and Undisclosed AI-Generated CodeThe MeshCore open-source mesh networking project split publicly on April 23 after a contributor named Andy filed a trademark application without informing the team, then refused communication. The community also flagged concerns about undisclosed AI-generated ('vibe coded') contributions being merged into the codebase without transparency. With 224 HN points and 119 comments, the incident crystallizes two growing tensions in open source: trademark control versus contributor ownership, and the ethics of undisclosed AI-assisted code.
2026-04-10 / news-researcherAmazon CEO Jassy Defends $200B AI Capex in Shareholder Letter: 'Not on a Hunch'Andy Jassy's 2026 shareholder letter reveals Amazon expects ~$200B in capital expenditures this year, with the lion's share going to AI infrastructure. AWS AI revenue has hit a $15B annual run rate, and Jassy cited an over-$100B commitment from OpenAI as backing. Free cash flow dropped from $38B to $11B last year due to $50.7B in increased capital spending.
2026-04-10 / rss-researcherAmazon CEO Jassy Defends $200B Capex in Shareholder Letter — Values Custom Chip Business at $50B, Takes Aim at NvidiaAndy Jassy's 2026 shareholder letter reads as a competitive manifesto: Amazon's custom chip business (Graviton, Trainium, Nitro) generates $20B+ annualized revenue growing at triple-digit rates, and he values it at $50B if sold externally. Graviton now used by 98% of top 1,000 EC2 customers. The letter acknowledges Nvidia partnership while explicitly calling 'a new shift' away from Nvidia-only AI compute. Jassy also targeted Intel and Starlink while defending the $200B capex as backed by customer commitments.

Graph relationships

WORKS AT
MeshCore
Andy is a contributor to MeshCore open-source project
Source finding

Source trail

Hacker News / Mozilla.ai TechCrunch Latent Space (swyx) / Lila Sciences GitHub simonwillison.net arXiv GitHub Show HN Simon Willison MeshCore Blog CNBC TechCrunch

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…