I tried building my startup entirely on European infrastructure
The article discusses the challenges encountered by a technology entrepreneur in starting a business in the European Union, including navigating complex regulations, finding talent, and securing funding. The author shares insights on the differences between launching a startup in the EU versus other regions.
MuMu Player (NetEase) silently runs 17 reconnaissance commands every 30 minutes
The article discusses the development of ChatGPT, an advanced language model created by OpenAI, and its potential impact on the field of AI. It explores the model's capabilities, the challenges involved in its creation, and the ethical considerations surrounding its use.
Consistency diffusion language models: Up to 14x faster, no quality loss
The article discusses the Consistency Diffusion, a new language model that aims to improve the consistency and coherence of text generated by large language models. The model is trained to maintain logical and factual consistency across multiple generated sentences, making the output more reliable and trustworthy.
Mystery donor gives Japanese city $3.6M in gold bars to fix water system
The article discusses the growing trend of professional gaming, or esports, and its increasing popularity and potential as a viable career path. It explores the rise of esports, the skills required, and the opportunities it presents for players, teams, and the industry as a whole.
Pi for Excel: AI sidebar add-in for Excel
This article discusses an open-source Python library called pi-for-excel that allows users to calculate the value of pi within Microsoft Excel. The library provides a user-friendly interface and supports various methods for calculating pi, making it a useful tool for educational and research purposes.
Reading the undocumented MEMS accelerometer on Apple Silicon MacBooks via iokit
The article discusses the reverse engineering of the accelerometer on Apple's M1 chip, including the discovery of its unique hardware and software implementation, and the implications for developers and researchers working with Apple's new silicon.
The path to ubiquitous AI (17k tokens/sec)
The article explores the advancements and challenges in making artificial intelligence (AI) ubiquitous, discussing the importance of improved training data, hardware capabilities, and ethical considerations to enable widespread adoption and integration of AI systems into various aspects of society.
Microsoft's new 10k-year data storage medium: glass
Microsoft has developed a new data storage medium made of glass that can store data for up to 10,000 years. This novel technology uses laser-etched digital data in quartz glass, providing a durable and long-lasting solution for preserving important information for future generations.
SwiftUI Agent Skill: Build Better Views with AI
The article explores the use of an AI-based SwiftUI agent to build better views. It discusses how the agent can help developers create more efficient and intuitive user interfaces by analyzing design patterns, identifying code smells, and suggesting improvements.
Web Components: The Framework-Free Renaissance
The article discusses the rise of web components as a framework-free approach to building web applications, highlighting their advantages in terms of modularity, browser-native integration, and the potential for a new era of web development without the need for large, opinionated frameworks.
Show HN: 17MB model beats human experts at pronunciation scoring
This article describes a Hugging Face Space that allows users to assess the pronunciation of their speech, providing feedback to improve language proficiency. The tool utilizes machine learning models to analyze audio recordings and offer detailed insights on pronunciation accuracy.
Microsoft Uses Plagiarized AI Slop Flowchart to Explain How Git Works
The article discusses Microsoft's use of a plagiarized AI-generated flowchart to explain how Git, a popular version control system, works. The article highlights the issues with relying on AI-generated content without proper attribution or fact-checking.
Show HN: Optimize_anything: A Universal API for Optimizing Any Text Parameter
We built optimize_anything, an API that optimizes any artifact representable as text — code, prompts, agent architectures, configs, even SVGs. It extends GEPA (our prompt optimizer, discussed here previously: https://arxiv.org/abs/2507.19457) far beyond prompts.
The API is deliberately minimal. You provide what to optimize and how to measure it:
import gepa.optimize_anything as oa
def evaluate(candidate: str) -> tuple[float, dict]: result = run_my_system(candidate) return result.score, {"error": result.stderr, "runtime": f"{result.time_ms}ms"}
result = oa.optimize_anything( seed_candidate="<your artifact>", evaluator=evaluate, )
The evaluator returns a score plus diagnostic feedback (we call it "Actionable Side Information" — stack traces, rendered images, profiler output, whatever helps diagnose failures). An LLM proposer reads this feedback during a reflection step and proposes targeted fixes, not blind mutations. Candidates are selected via a Pareto frontier across metrics/examples, so a candidate that's best at one thing survives even if its average is mediocre.
Two ideas distinguish this from AlphaEvolve/OpenEvolve/ShinkaEvolve-style LLM evolution: (1) diagnostic feedback is a first-class API concept rather than a framework-specific mechanism, and (2) the API unifies three optimization modes — single-task search (solve one hard problem), multi-task search (solve related problems with cross-transfer), and generalization (build artifacts that transfer to unseen inputs). Prior frameworks only express mode 1.
We tested across 8 domains. Selected results:
Coding agent skills: Learned repo-specific skills push Claude Code to near-perfect task completion and make it 47% faster Cloud scheduling: Discovered algorithms that cut costs 40%, topping the ADRS leaderboard over expert heuristics and other LLM-evolution frameworks Agent architecture: Evolved a 10-line stub into a 300+ line ARC-AGI agent, improving Gemini Flash from 32.5% → 89.5% Circle packing (n=26): Outperforms AlphaEvolve's published solution Blackbox optimization: Generated problem-specific solvers matching or exceeding Optuna across 56 EvalSet problems CUDA kernels: 87% match or beat baseline; multi-task mode outperforms dedicated single-task runs
``` pip install gepa ```
Blog with full results and runnable code for all 8 case studies: https://gepa-ai.github.io/gepa/blog/2026/02/18/introducing-o...
GitHub: https://github.com/gepa-ai/gepa
Show HN: Fostrom, an IoT Cloud Platform built for developers
Hey HN! Arjun and Sid here.
Fostrom is an IoT Cloud Platform designed for developers to make it really easy to get started and scale fleets. We have Device SDKs (in Python, JS, Elixir, more coming soon), Typed Schemas, Per-Device Mailboxes, Programmable Actions, 4 Global Regions for lower-latency connections, and much more.
We've built Fostrom to solve a real need we faced in our previous startup, building a fully automated indoor vertical farm. We spent more time figuring out IoT infrastructure than writing automation logic. Fostrom is the platform we wished existed back then.
Over the last few years we've experimented with a lot of interesting tech and architectures, and settled on an architecture that we believe is quite elegant. We wrote a Go<->Elixir bridge to execute JS code in WASM for Actions, implemented a DuckDB library for Elixir, and wrote a Device Agent in Rust that our SDKs run in the background (https://github.com/fostrom/devicekit).
The most interesting realization we had was about the data architecture. For years, we tried using distributed databases and built complex layers on top of them, but all approaches had significant limitations specifically around consistency and querying. We want to provide operational correctness, rich insights, and reliability. Finally, we came to the conclusion that to achieve this we really need a SQL database for fleet data. So we built a DuckDB-based replicated multi-tenant data layer. We're still improving it (hence the Technical Preview badge) but we're quite proud of this decision. It simplifies the rest of the codebase, while keeping operational complexity in just a few places.
Our vision is to make a powerful IoT platform that enables you to build correct, secure, and reliable connected systems without dealing with any of the plumbing or infrastructure.
Next up, we're gonna launch our CLI, add automatic device monitoring to our Device SDKs, and improve the debugging experience. We have some pretty cool ideas to make Fostrom and the experience of developing connected systems better. We're also going to write more about our architecture and journey soon.
We also published our launch blog post which goes into more detail about our vision, what we've built, and our future plans: https://fostrom.io/blog/introducing-fostrom
Would love for you to try out Fostrom and give us your feedback and thoughts.
A Primer on Terrorism, Insurgency and Protest
This article provides a primer on the complex topics of terrorism, insurgency, and protest, exploring the strategies, tactics, and motivations of groups that challenge state authority through non-conventional means. It aims to offer a nuanced understanding of these phenomena and the factors that contribute to their emergence and evolution.
Minions – Stripe's Coding Agents Part 2
The article discusses the development of an end-to-end coding agent, Minions Stripes, which is designed to automate the process of creating and managing Stripe-powered applications. It highlights the agent's capabilities, such as handling payments, managing customer data, and integrating with other services.
Internationalise the Fediverse
The article discusses the importance of internationalization in the Fediverse, a decentralized social network. It highlights the need for better support for non-English languages and cultures, and the challenges involved in achieving this goal across different Fediverse platforms.
FBI 'Asset' Helped Run a Dark Web Site That Sold Fentanyl-Laced Drugs for Years
The article reveals that an FBI informant helped run a dark web site that sold fentanyl-laced drugs for years, raising questions about the bureau's role in enabling the sale of dangerous narcotics.
Archive.today CAPTCHA page executes DDoS; Wikipedia considers banning site
Wikipedia is considering blacklisting the Archive.today website after its maintainer allegedly DDoSed a blog critical of the site. The article discusses the ongoing dispute between the two platforms and the potential consequences for Archive.today.
Judge scolds Mark Zuckerberg's team for wearing Meta glasses to trial
Meta, the parent company of Facebook, is embroiled in a trial over its attempt to acquire a startup that develops AI-powered smart glasses. The trial focuses on whether the acquisition would violate antitrust laws and stifle competition in the nascent smart glasses market.