AI News

Daily AI News for Builders

The stories worth reading, for builders and indie hackers. Updated all day.

Agentic AI Projects Face Grim Survival Odds by 2028

Gartner predicts that more than 40% of agentic AI projects won't survive until 2028 due to escalating costs, unclear business value, and inadequate risk controls. This is a wake-up call for developers and teams relying on these technologies. Only about 30% of organizations have reached a governance maturity level of three or higher in handling agentic AI risks. The race now is not just about building the most capable agents but ensuring they're trustworthy and controllable. Key numbers: Over 40% failure rate by 2028, only 30% mature in governance.

Aug 22, 2026 · 3 min read

DOJ Investigates a16z for Board Seats With Rival AI Firms

The DOJ is probing venture firm Andreessen Horowitz (a16z) over its board seat arrangements with competing AI startups. This investigation, spanning nearly a year, hints at deeper issues beyond simple conflict-of-interest concerns. For developers and startup founders, this could mean stricter scrutiny on VC ties and potential policy changes affecting the tech ecosystem.

Aug 22, 2026 · 3 min read

tech

Programming Languages Matter Less as LLMs Take Over

The idea that 'programming is solved now' has gone viral on Twitter as LLMs like Claude can rewrite code in any language without needing deep familiarity with it. This shift means developers are choosing languages based more on marketing hype than technical merit, leading to a rise in Rust and Zig adoption due to their ability to produce fast software. Cloudflare's Artifacts service uses a pure-Zig Git-protocol engine compiled into a 100 KB WebAssembly module, highlighting the trend towards smaller and faster technologies.

Aug 22, 2026 · 3 min read

tech

NetBSD Powers UK Company's Servers With 29 Machines

A UK-based company has transitioned its 29 high-end servers to run on NetBSD, marking a shift from their previous setup of Windows on older hardware. This move was driven by the author's negative experience with Windows administration and a smooth trial period at home. The switch resulted in improved stability and efficiency for MySQL databases, Apache web services, Postfix email handling, and Samba file sharing. With NetBSD managing over 870GB of data daily and dealing with up to 35 HTTP requests per minute during peak hours, the company's IT infrastructure has become more reliable and cost-effective.

Aug 22, 2026 · 3 min read

Faraday AI Agent Outperforms Giants on Paper Replication

Inherent, a London startup founded by ex-DeepMind alumni, unveiled its AI agent Faraday, which outperforms much larger models in independently reproducing the findings of published scientific papers. The kicker? Faraday does this with just 27 billion parameters, compared to competitors' hundreds of billions. This matters for researchers and developers looking to automate complex tasks without bloated models. Inherent's model uses reinforcement learning to develop 'research taste,' identifying valuable experiments autonomously. With a $50 million seed round under its belt, the company is set to grow from 12 to around 20-25 employees by year-end.

Aug 22, 2026 · 3 min read

tech

Company Launches HDR Tech for Design Elements

The company introduced a tech that auto-selects whites for design elements like wordmarks and orbit marks. This is huge for designers looking to streamline their workflow without compromising on aesthetics or quality. For instance, the technology ensures the white ring in an orbit mark stands out against its brand color #2f6bff by pushing both colors to the same peak brightness level. The HDR format enhances JPEG images, making them more vibrant and detailed.

Aug 22, 2026 · 3 min read

The daily that does your AI homework.

One email, 7am, free. Reads in 5 minutes.

tech

LLM Benchmarks: What Home Lab Users Need to Know

Local language model (LLM) performance varies widely based on hardware generation and benchmark type. Zero-shot tests fail to capture real-world performance for agentic tasks. Running comprehensive benchmarks, including long-context tool-calling and domain-specific knowledge evaluations, is essential. This ensures accurate assessments of an LLM's setup weaknesses. Key considerations include correct sampler settings from model cards, temperature adjustments, KLD divergence direction, vocabulary truncation impact, and evaluation text calibration data.

Aug 22, 2026 · 3 min read

tech

Thinking in Python: Comprehensive Guide With 44 Chapters

The book 'Thinking in Python' is a comprehensive guide divided into five parts: Introduction, Foundations, Techniques, Patterns, and Functional Programming. With 44 chapters, it covers core concepts like containers, control flow, functions, modules, classes, static typing, testing, data classes, pattern matching, decorators, context managers, comprehensions, metaprogramming, rethinking objects, iterators, singleton patterns, and more. The book is freely readable online but requires permission for reproduction or distribution. It's a must-read for developers looking to deepen their understanding of Python’s advanced features and best practices.

Aug 22, 2026 · 3 min read

tech

Author's Pittsburgh Move: Winter Renovation Struggles

The author moved into a house in Pittsburgh during winter with no working utilities or heating, leading to a series of renovation challenges. January was particularly tough as the author had to install gas lines while dealing with frozen pipes. Scrap collectors took advantage of the situation by stealing valuable items like cast-iron bathtubs and ductwork from the dumpster. The furnace, weighing 17 steps up to the first floor, required significant effort to move due to narrow doorways and claustrophobia issues.

Aug 22, 2026 · 3 min read

tech

Belgian Prince's Son Confirmed via DNA Test

DNA testing revealed Clément Vandenkerckhove is indeed the biological son of Belgium's Prince Laurent. This confirmation came after years of speculation about the paternity of a child born to Wendy Van Wanten, who dated Prince Laurent in the '90s. The test result solidifies Clément’s legal standing within the royal family, granting him equal inheritance rights to his half-siblings from Laurent's private estate but excludes him from official duties and the throne. This story highlights how long-standing secrets can be resolved through modern technology.

Aug 22, 2026 · 3 min read

tech

Old TV Becomes Free Monitor for PC

Turn that dusty old TV into a free computer monitor by simply connecting it via HDMI cable or using wireless screen mirroring like Apple's AirPlay or Google Cast. Why? It’s an easy way to get more screen real estate for less money. For instance, a 32-inch Hisense A4 TV costs around $110 and offers 4K resolution with DTS Virtual:X audio. However, it tops out at 60Hz refresh rate, which might cause frame rate issues in demanding games.

Aug 22, 2026 · 3 min read

tech

Anthropic Tweaks Claude Effort Scale Without Warning

Starting August 22, Anthropic quietly adjusted how its Claude AI model reads user input efforts without warning users. This tweak could lead to unexpected behavior for apps relying on previous interpretations of effort levels. The change affects version 2.1.237 and later, where 'high' effort is now rated as 10 out of 100, previously equivalent to what was marked as 'low'. Users reported issues with their applications due to this sudden shift in how Claude processes input data.

Aug 22, 2026 · 3 min read

Page 1 of 695