cuppa

brewed 03:10 AM
← Jul 12

Monday

jul13

2026

Jul 14 →

the brief

Today split between incremental but meaningful dev tooling and cost‑efficiency realities. Next.js pushes Request Insights forward while Cursor reportedly broadens into a general agent, and Anthropic keeps Fable 5 live a bit longer. On the research side, Rust gets a model checker and Google experiments with AI‑assisted peer review, as Apple’s next chips and rising power costs foreshadow the infrastructure battles ahead.

the poursit · sip · 11 items

pulse

(05)
  • vercel/next.js· feedJul 12, 11:57 PM

    Next.js canary adds request insights

    Canary v16.3.0 advances the new Request Insights with dev snapshots, request history derivation, and local framework spans, improving HMR tooling visibility for debugging and performance.

    v16.3.0-canary.84 — Misc Changes request insights: expose dev snapshots to tools and HMR (3/5): #93976 request insights: derive request history and fetch data (2/5): #93975 request insights: record local framework spans (1/5): #93974 Credits Huge thanks to @feedthejim for helping!

  • techmeme· feedJul 12, 10:30 PM

    Cursor building general‑purpose agent Sand

    The Information reports Cursor is developing “Sand,” an email/text/doc agent for non‑developers, positioning the code editor startup to challenge Claude Cowork with broader, everyday workflows.

    Sources: Cursor is building a general-purpose AI agent codenamed Sand, aimed at non-developers, that handles emails, texts, and documents to rival Claude Cowork (Grace Kay/The Information) — Grace Kay / The Information: Sources: Cursor is building a general-purpose AI agent codenamed Sand, aimed at non-developers, that handles emails, texts, and documents to rival Claude Cowork — Cursor is developing a general-purpose AI agent meant to compete with popular tools like Anthropic's Claude Cowork …

    signal 5hype 2agentproduct_rumorcompetitive_landscapesource ↗
  • simonw/blog· feedJul 12, 09:20 PM

    Anthropic extends Fable 5 availability

    Claude Fable 5 remains on all paid plans through July 19, with Claude Code weekly limits 50% higher—keeping frontier access and coding capacity during the GPT‑5.6 Sol rollout.

    Fable gets another bump — <p>One of the consequences of GPT-5.6 Sol being clearly a Fable/Mythos class model is that Anthropic have, once again, <a href="https://x.com/claudeai/status/2076351399999557669">bumped the date</a> that Fable stops being available in their Claude Max plans:</p> <blockquote> <p>We're extending Claude Fable 5 access on all paid plans, as well as keeping Claude Code’s weekly rate limits 50% higher, through July 19.</p> <p>As before, you can use up to half of your weekl...

    signal 8hype 1anthropicclaude_fableclaude_codesource ↗
  • simonw/blog· feedJul 12, 11:46 PM

    shot‑scraper 1.11 improves server timing

    The CLI’s server: mechanism now works even if startup exceeds a second across video and multi modes, plus option consistency polish—smoother scripted capture pipelines.

    shot-scraper 1.11 — <p><strong>Release:</strong> <a href="https://github.com/simonw/shot-scraper/releases/tag/1.11">shot-scraper 1.11</a></p> <p>Some minor improvements, mainly around command option consistency and making the <code>server:</code> mechanism <a href="https://shot-scraper.datasette.io/en/stable/multi.html#running-a-server-for-the-duration-of-the-session">used by</a> both <code>shot-scraper video</code> and <code>shot-scraper multi</code> work if the server takes longer than a se...

    signal 6hype 1release_notescli_toolweb_automationsource ↗
  • techmeme· feedJul 12, 03:20 PM

    Apple M7 rumored with bigger NPU

    Bloomberg says Apple is taping out M7 with major NPU upgrades and planning M7 Ultra with up to 1.5TB RAM—telegraphing heavier on‑device AI workloads ahead.

    Sources: Apple is already taping out M7 with major NPU upgrades, and plans M7 Ultra with 1.5TB RAM, and M8, for 2028; new Pencils are coming with a new iPad Pro (Mark Gurman/Bloomberg) — Mark Gurman / Bloomberg: Sources: Apple is already taping out M7 with major NPU upgrades, and plans M7 Ultra with 1.5TB RAM, and M8, for 2028; new Pencils are coming with a new iPad Pro — Also: New Apple Pencils are coming. — Apple's new Mac chip road map represents the company's latest move to rebuild …

findings

(04)
  • paper.bsky.socialJul 13, 12:06 AM

    Kani brings model checking to Rust

    From AWS and collaborators, Kani symbolically executes Rust to automatically prove memory safety and functional properties on critical code, extending formal methods into mainstream systems programming.

    [17/30] 160 Likes, 9 Comments, 1 Posts 2607.01504, cs․SE | cs․LO | cs․PL, 01 Jul 2026 🆕Kani: A Model Checker for Rust Rémi Delmas, Zyad Hassan, Qinheping Hu, Rahul Kumar, Felipe R. Monteiro, Thanh Nguyen, Adrián Palacios, Celina Val, Michael Tautschnig, Justus Adam, Daniel Schwartz-Narbon...

    signal 6hype 1paperrustformal_verificationsource ↗
  • paper.bsky.socialJul 13, 12:06 AM

    Automating peer review with Paper Assistant

    Google researchers detail a tool that drafts structured scientific reviews from papers, exploring rubric design, quality controls, and human‑AI collaboration for scalable, transparent scholarly evaluation.

    [20/30] 153 Likes, 29 Comments, 5 Posts 2606.28277, cs․LG | cs․AI | cs․CL | cs․CY, 26 Jun 2026 🆕Towards Automating Scientific Review with Google's Paper Assistant Tool Rajesh Jayaram, Drew Tyler, David Woodruff, Corinna Cortes, Yossi Matias, Vahab Mirrokni, Vincent Cohen-Addad

    signal 5hype 2paper_releasegoogleresearchsource ↗
  • paper.bsky.socialJul 13, 12:05 AM

    Data centers raise local power bills

    New causal study finds US data center build‑outs measurably increase residential electricity prices, quantifying an externality that will shape AI infra siting and regulatory debates.

    [21/30] 143 Likes, 215 Comments, 3 Posts 2606.19777, physics․soc-ph | econ․GN, 18 Jun 2026 🆕Have Data Centers Raised Your Electric Bill? Causal Evidence from the United States Asa Watten, John Bistline, Geoffrey Blanford

    signal 4hype 1paperdata_centersenergy_economicssource ↗
  • tmlr-pub.bsky.socialJul 13, 12:20 AM

    Sparse tensor sharing for transformers

    A TMLR paper learns fine‑grained parameter sharing through sparse tensor decomposition, aiming to cut model size and compute while preserving accuracy—another path toward efficient LLMs.

    Learning Fine-grained Parameter Sharing via Sparse Tensor Decomposition Cem Üyük, Mike Lasby, Mohamed Yassin, Utku Evci, Yani Ioannou Action editor: Yoshitomo Matsubara https://openreview.net/forum?id=vbS7Z8Zswe #transformer #sparse #tensor

    signal 6hype 1papersparsityparameter_sharingsource ↗

voices

(02)
  • techmeme· feedJul 12, 09:40 PM

    Cost pressure reshapes AI competition

    Bloomberg argues OpenAI, Meta, and SpaceXAI are pushing cheaper inference, putting Anthropic under margin pressure as enterprise buyers scrutinize AI ROI and vendor pricing power.

    OpenAI, Meta, and SpaceXAI may be able to put pressure on Anthropic by emphasizing cost efficiency, as business customers increasingly scrutinize AI spending (Bloomberg) — Bloomberg: OpenAI, Meta, and SpaceXAI may be able to put pressure on Anthropic by emphasizing cost efficiency, as business customers increasingly scrutinize AI spending — Three prominent artificial intelligence developers released new models over the past week. They all promise to be more advanced …

    signal 5hype 2market_competitionpricingenterprise_aisource ↗
  • simonw/blog· feedJul 12, 11:57 PM

    On defining the DRI role

    Simon Willison surfaces GitLab’s crisp definition of “Directly Responsible Individual”—a useful pattern for assigning ownership in fast‑moving AI and tooling projects.

    Directly Responsible Individuals (DRI) — <p><strong><a href="https://handbook.gitlab.com/handbook/people-group/directly-responsible-individuals/">Directly Responsible Individuals (DRI)</a></strong></p> I went looking for a definition of "Directly Responsible Individuals" and the best I found was in the GitLab handbook. Apparently the term originated at Apple, where it's used to describe the person who is "ultimately accountable for the success or failure of a specific project, initiative, or ...

    signal 5hype 1managementprocessleadershipsource ↗