GEEK HAUS
← All dates

Daily reads: Jul 25, 2026

12 reads selected from 17 stories across 6 sources published in this period.

We take turns across sources, starting with their most recent stories. This is a reading selection, not an importance ranking or an archive of the email newsletter.

Dates follow Korea Standard Time. Weekly selections cover Monday through Sunday.

Our editorial policy and selection method
  1. 01 / The Verge

    Warner Bros. is suing Amazon for poaching employees

    Warner Bros. Discovery filed a lawsuit accusing Amazon of an unlawful employee poaching campaign, including hiring former originals marketing senior VP Pia Barlow. The suit says Barlow remains under contract until October 31, while reports suggest HBO drama chief Francesca Orsi may also be an Amazon target.

    Why it matters

    The dispute highlights intensifying competition for Hollywood executive talent as streaming platforms fight to strengthen originals, marketing, and production leadership. If Warner prevails, it could reinforce stricter enforcement of executive contracts and slow aggressive hiring moves across the entertainment industry.

  2. 02 / TechCrunch

    SpaceX launches new V3 Starlink satellites but suffers another booster failure

    SpaceX completed several objectives during its second Starship V3 flight, including launching new V3 Starlink satellites. However, the mission appears to have suffered another booster failure tied to relighting the rocket engines.

    Why it matters

    The mixed result shows SpaceX is making progress on payload deployment while still facing reliability challenges in Starship’s recovery architecture. Solving engine relight issues will be critical for reusable operations and the company’s broader launch cadence ambitions.

  3. 03 / artificialanalysis.ai

    Claude Opus 5 tops Artificial Analysis intelligence rankings, while rivals lead on speed, latency, cost, and context size

    Artificial Analysis ranks Claude Opus 5 variants as the highest-scoring models on its Intelligence Index, ahead of Claude Fable 5 and GPT-5.6 Sol. The leaderboard also highlights separate leaders for output speed, latency, price, and context window, showing that no single model dominates every metric.

    Why it matters

    The results underscore a growing split between intelligence benchmarks and operational factors like cost and speed. As AI buyers compare models, leaderboards that combine quality, price, latency, and context may increasingly shape procurement and developer adoption decisions.

  4. 04 / VentureBeat

    VentureBeat Research: Where enterprise AI agent governance hasn't caught up

    VentureBeat Research surveyed enterprises across five AI agent control layers: identity, evaluation, cost telemetry, context, and orchestration. It found many companies deployed agents before governance was mature, with 57% to 68% planning to switch or add vendors within 12 months, while most so-called AI agents are still closer to chatbots than autonomous systems.

    Why it matters

    The findings suggest enterprise AI adoption is moving from experimentation to a harder phase focused on governance infrastructure. As agents gain autonomy over workflows such as production code or system changes, buyers are likely to prioritize trust, monitoring, and control planes over standalone chatbot capabilities.

  5. 05 / dbos.dev

    Postgres LISTEN/NOTIFY can scale to 60K writes per second with careful stream design and tuning

    The article argues that Postgres LISTEN/NOTIFY is often dismissed unfairly as unscalable due to misunderstood performance behavior around a global lock. The authors describe using it for low-latency streams and pub/sub, ultimately reaching 60K writes per second on a single Postgres server with millisecond-scale latency.

    Why it matters

    This suggests Postgres can handle more real-time application patterns than many teams assume, reducing the need for separate messaging systems in some architectures. The broader trend is toward simplifying infrastructure by pushing streaming workloads into databases when latency, durability, and operational tradeoffs are acceptable.

  6. 06 / anthropic.com

    Anthropic publishes the system card for Claude Opus 5, outlining its capabilities, safety testing, and deployment details

    Anthropic has released a system card for Claude Opus 5, offering documentation on the model’s behavior, evaluations, and safeguards. The linked document appears to serve as the primary source for technical and safety information about the new model.

    Why it matters

    System cards are becoming a key transparency tool as frontier AI labs face growing scrutiny from users, enterprises, and regulators. The release signals continued emphasis on AI safety documentation as competitive model launches become more frequent and consequential.

  7. 07 / The Verge

    Is this e-reader case a gun?

    Louisville creator Luke Blackford showed off the “Gundle,” a 3D-printed case that makes the minimalist Xteink X4 e-reader look like a handgun. The accessory is modeled after a real pistol frame but cannot fire, and its trigger has no function.

    Why it matters

    The design highlights how 3D printing can blur the line between novelty accessories and objects that may be perceived as threats in public. Even harmless gadgets shaped like weapons could raise safety, moderation, and platform-policy questions as maker culture becomes more visible online.

  8. 08 / TechCrunch

    Prentis, new AI lab co-founded by Reid Hoffman, Mark Pincus in talks to raise $100M

    Prentis, a new AI lab focused on computer-use models, is reportedly in talks to raise $100 million at a $1 billion valuation. The company says it is building AI agents that can automate office workflows across documents and business systems, and has signed customer contracts tied to performance-based savings.

    Why it matters

    The raise signals continued investor appetite for startups moving beyond chatbots toward agents that can complete real workplace tasks. If Prentis can deliver lower-cost automation reliably, AI agents may become more commercially viable in back-office industries like insurance, healthcare, manufacturing, and logistics.

  9. 09 / VentureBeat

    Anthropic launches Claude Opus 5, a cheaper AI model for coding, agents and enterprise workflows

    Anthropic released Claude Opus 5, saying it offers nearly the intelligence of its top Claude Fable 5 model at about half the cost. The model is available across Anthropic’s platforms, becomes the default for Claude Max, and is positioned for coding and enterprise workflows that need strong performance without frontier-model pricing.

    Why it matters

    The launch reflects a shift in AI competition from maximizing benchmark capability to optimizing the cost-performance sweet spot for everyday business use. For enterprises, models like Opus 5 could accelerate adoption by making agentic coding and knowledge-work tasks economically practical at larger scale.

  10. 10 / The Verge

    After backlash, Meta pauses plan to ‘rate limit’ its smart glasses

    Meta has paused plans to put Conversation Focus, a smart glasses feature that helps users hear nearby speech more clearly, behind a $20 monthly subscription. The company says the feature will remain free for early testers through its Early Access Program, though premium features are still expected to become subscription-based over time.

    Why it matters

    The reversal shows how sensitive users are to subscriptions for features that feel like core functionality, especially accessibility-related tools. Meta still appears committed to recurring revenue from wearables, suggesting future smart glasses may increasingly split capabilities into free and paid tiers.

  11. 11 / TechCrunch

    TechCrunch Disrupt 2026’s new Smart Money Stage explores fintech, payments, AI, and everything between

    TechCrunch Disrupt 2026 will feature a new Smart Money Stage dedicated to how money is changing beyond cash and traditional bank accounts. The programming will explore fintech, payments, AI, and related areas shaping financial services.

    Why it matters

    The new stage reflects how financial technology has become a core startup and investor theme rather than a niche category. Its focus on AI-driven finance points to growing interest in automation, personalization, and infrastructure changes across payments and banking.

  12. 12 / The Verge

    Meta just created a moderation nightmare for its smart glasses

    Meta’s smart glasses are drawing criticism as users film strangers without clear consent, including prank content and men recording interactions with women. Business Insider reports Meta has begun banning some glasses-shot content, underscoring the company’s struggle to manage misuse of the device.

    Why it matters

    The controversy shows how wearable cameras can turn everyday public spaces into arenas for unsolicited content and harassment. Meta’s response may set an early precedent for how platforms police media captured by always-available AI and camera-enabled devices.