~/bookmarks

agent-native, drop-in Github Actions that run locally or self-hosted in microvms, with debug/pause-on-failure and step-level retries - preloopdev/preloop

Ultramarine 1 Flash is an open model for professional work. It makes the decks, models, and memos that people in finance, law, consulting, and operations put in front of clients, at a fraction of the cost of frontier models.

Search open government tenders, RFPs and public procurement contracts from 100+ countries in one place. Every notice in English, updated daily, with a link to the official portal to bid.

MCP server and CLI that crawls websites for AI agents: finds the answer page 2x more often than Crawl4AI, 10x cheaper per answer than letting Claude crawl. - RayenMlayeh/jevcrawl

The Hacker News thread discusses why multiple AI frontier companies reduce the risk of concentrated power from recursive self-improvement. Commenters identify verifying successive AI versions as a core difficulty, noting both humans and prior versions lack the capacity to evaluate newer systems. The discussion covers intelligence scaling metrics and proposes market competition and multi-environment training as approaches to recursive self-improvement.

A whole world made of bricks, shared with everyone online. Race, fly, ride the bus, build your own place – free in your browser, on your phone or your computer.

Give every customer an agent in its own isolated sandbox, at up to 95% lower cost than most sandbox providers.

Let an LLM generate the questions to ask Jev for a given query. See how well that works as a reranker.

Fine-tune LLMs like Qwen and Gemma to make decisions with calibrated probabilities.

i built @WisprFlow’s paid ads program here’s how i’d rebuild it in 30 days

Agents are unlocking enormous productivity gains for customers, but their ability to work across files, networks, and applications can introduce new security risks. This can leave customers feeling like they only have two choices: give agents unrestr

A curated list of 111 startup directories and launch platforms with Domain Rating, traffic and pricing for each. Find where to submit your startup for users, feedback and backlinks.

Rebalancer is a domain-specific language and tool for specifying and solving assignment problems (eg, putting balls in boxes with complex rules). - facebook/rebalancer

A curated, source-backed list of projects built with Jev, TypeSafe AI’s System One model for typed decisions. - cobanov/awesome-jev

An awesome collection of Jev use cases, workflows, and agent skills. - wuyoscar/jev-skill

interfaze-1-lite is the first open-weight model built for deterministic work like OCR, speech-to-text and classification, with typed output, boxes and confidence scores on a single 80 GB GPU.

Multiplatform client app used to browse buckets and file in Supabase project - hieuwu/supabase-storage-client

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Hacker News discussion covers running the 125B Qwen 3.8 Flash Next model on consumer GPUs (RTX 3090, 4090, 5090) using 3-bit to 5-bit quantization, achieving 40 to 125 tokens per second. Benchmarks show IQ3_XXS quantization scoring 89.1% on code tasks versus 71.9% for a Q4/Q5 mix, with users reporting fewer hallucinations than 27B 4-bit variants. Initial setup takes about one hour, cold model loading into GPU memory takes roughly 15 minutes, and users compare $1/hour GPU rental costs against $20/month API subscriptions.

Qwen3.8-Flash-Next (125B MoE) on a 8GB+ NVIDIA GPU: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input. - Niko1221/Strata

An AI assistant that lives in your browser. Built for collaboration, not autonomy theater. You guide, it executes. Automate repetitive web tasks, extract data from any website, and transform it int…

Observation, targets, the background-first action ladder, verification, and who owns the runtime.

Lightpanda leaves beta with 1.7 million passing Web Platform Tests and CORS on by default. Here’s what changed since November 2024.

The former Steelers coach went deep sharing his passion project on YouTube, and teased more to come.