June 29, 2026
Report summary
5 stories cleared the bar, led by Interactions API, HackerRank open sourced its ATS. My resume scored 90/100. Oh wait 74. No – 88, and GLM 5.2 beats Claude in our benchmarks.
Worth attention
Google says the Interactions API is now the default path for new Gemini apps, with server-side state, background execution, and better cache economics.
A reproducible writeup shows HackerRank's open ATS can swing resume scores widely across identical runs, which makes LLM-only screening look too noisy for hard cutoffs.
Semgrep reports GLM 5.2 beat Claude Code on its IDOR benchmark under a minimal prompt harness, but the claim is vendor-authored and limited to one task.
EFF argues the proposed KIDS Act would pressure platforms into broad age checks and more restrictive moderation, especially around private messaging.
Aleph Alpha describes an internal “model training as code” stack that treats training pipelines like CI-driven software, with immutable artifacts and nightly end-to-end validation.
Full digest
Word-game Show HN, but not relevant to the memo's builder decision filter.
Interesting DRM directory project, but outside the memo's core software/operator focus.
Tiny exit anecdote without enough transferable detail to justify inclusion.
Meme-style post, not a substantive builder report.
Builder says 600 trial video generations did not convert once a $6 paywall was added, pointing to a value-timing problem more than demand proof.
First-sale milestone post, but too personal and thin for the memo.
A founder selling into offline buyers reports cold calls and tailored Instagram outreach work better than standard online channels.
Trailer-video feedback request, not decision-relevant.
Open-ended discussion prompt, not evidence of a change worth acting on.
Growth stall post in an adult niche, but too specific and anecdotal to elevate.
$1k MRR anecdote without enough hard detail to beat stronger items.
Devtool distribution question, interesting but still mostly request-for-advice noise.
First customer milestone without broader lesson or evidence.
Pre-registration launch tactics post, but too speculative.
Pricing thought experiment without evidence beyond opinion.
Question about QA for AI-built client apps, but the linked Reddit page was blocked during browsing and the item lacks enough signal on its own.
Video production advice request, not a memo item.
Closed-beta discussion prompt, still too early-stage and generic.
Generic MVP advice post, not new enough to keep.
Complaint about repeating AI context, but no concrete method or product change.
Beta launch self-promo, too thin.
Affiliate platform question, not evidence.
Hiring request for video editing help, not memo material.
Salon wait-time idea pitch, still at problem-statement stage.
Travel app promo and dev question, outside scope.
Google says the Interactions API is now the default path for new Gemini apps, with server-side state, background execution, and better cache economics.
A reproducible writeup shows HackerRank's open ATS can swing resume scores widely across identical runs, which makes LLM-only screening look too noisy for hard cutoffs.
Semgrep reports GLM 5.2 beat Claude Code on its IDOR benchmark under a minimal prompt harness, but the claim is vendor-authored and limited to one task.
Opinion essay on age verification, but the EFF piece carried the stronger practical signal.
Memory price history is interesting reference data, not tomorrow-morning action.
Historical menu collection project, off-topic for this memo.
Medical self-use story with risky framing, not something to amplify.
Older black-box LLM distillation paper resurfaced on HN, but not a fresh development.
Language-history post, outside scope.
Duplicate HN entry for Zanagrams.
Essay on token economics and agent tooling, but not concrete enough versus stronger workflow items.
Old Bloomberg feature recirculating, not actionable.
EFF argues the proposed KIDS Act would pressure platforms into broad age checks and more restrictive moderation, especially around private messaging.
Aleph Alpha describes an internal “model training as code” stack that treats training pipelines like CI-driven software, with immutable artifacts and nightly end-to-end validation.
AI cheating in higher ed is real, but this item does not change builder decisions here.
Original markdown
# Nightly Librarian — Newsletter draft Run: 0e078be6-85e1-461e-9349-b03a40926651 Started: 2026-06-29T11:02:38.527Z Completed: 2026-06-29T11:07:20.870Z ## Worth attention - **Interactions API** https://ai.google.dev/gemini-api/docs/interactions-overview?hl=vi Google says the Interactions API is now the default path for new Gemini apps, with server-side state, background execution, and better cache economics. - **HackerRank open sourced its ATS. My resume scored 90/100. Oh wait 74. No – 88** https://danunparsed.com/p/hackerrank-open-source-ats A reproducible writeup shows HackerRank's open ATS can swing resume scores widely across identical runs, which makes LLM-only screening look too noisy for hard cutoffs. - **GLM 5.2 beats Claude in our benchmarks** https://semgrep.dev/blog/2026/we-have-mythos-at-home-glm-52-beats-claude-in-our-cyber-benchmarks/ Semgrep reports GLM 5.2 beat Claude Code on its IDOR benchmark under a minimal prompt harness, but the claim is vendor-authored and limited to one task. - **The KIDS Act would require age checks to get online** https://www.eff.org/deeplinks/2026/06/kids-act-would-require-age-checks-get-online EFF argues the proposed KIDS Act would pressure platforms into broad age checks and more restrictive moderation, especially around private messaging. - **Model Training as Code** https://aleph-alpha.com/en/blog/model-training-as-code/ Aleph Alpha describes an internal “model training as code” stack that treats training pipelines like CI-driven software, with immutable artifacts and nightly end-to-end validation. ## Full digest - [R] [hn-show] Show HN: Zanagrams — https://zanagrams.com/ — Word-game Show HN, but not relevant to the memo's builder decision filter. - [R] [hn-show] Show HN: DRM-Free Books — https://frequal.com/Perspectives/DrmFreeAuthors.html — Interesting DRM directory project, but outside the memo's core software/operator focus. - [R] [reddit-saas] I sold my $29 mrr SaaS for 1.4k - AMA — https://www.reddit.com/r/SaaS/comments/1uiac7n/i_sold_my_29_mrr_saas_for_14k_ama/ — Tiny exit anecdote without enough transferable detail to justify inclusion. - [R] [reddit-saas] Hope Driven Development (HDD) — https://www.reddit.com/r/SaaS/comments/1ui4eq2/hope_driven_development_hdd/ — Meme-style post, not a substantive builder report. - [R] [reddit-saas] Got 600 videos generated from my tool, but can't get anyone to pay — https://www.reddit.com/r/SaaS/comments/1uihxer/got_600_videos_generated_from_my_tool_but_cant/ — Builder says 600 trial video generations did not convert once a $6 paywall was added, pointing to a value-timing problem more than demand proof. - [R] [reddit-saas] My first sale after a year of absolute silence — https://www.reddit.com/r/SaaS/comments/1uij271/my_first_sale_after_a_year_of_absolute_silence/ — First-sale milestone post, but too personal and thin for the memo. - [R] [reddit-saas] How do you get your first 10 customers when your buyers don't hang out online? — https://www.reddit.com/r/SaaS/comments/1uikeof/how_do_you_get_your_first_10_customers_when_your/ — A founder selling into offline buyers reports cold calls and tailored Instagram outreach work better than standard online channels. - [R] [reddit-saas] How is this trailer video?? — https://www.reddit.com/r/SaaS/comments/1uhxvkd/how_is_this_trailer_video/ — Trailer-video feedback request, not decision-relevant. - [R] [reddit-saas] If a SaaS company suddenly had an extra $100k, what problem would you actually spend it on? — https://www.reddit.com/r/SaaS/comments/1uik2fc/if_a_saas_company_suddenly_had_an_extra_100k_what/ — Open-ended discussion prompt, not evidence of a change worth acting on. - [R] [reddit-saas] What am I doing wrong ? How can I reach users ? — https://www.reddit.com/r/SaaS/comments/1uif2sy/what_am_i_doing_wrong_how_can_i_reach_users/ — Growth stall post in an adult niche, but too specific and anecdotal to elevate. - [R] [reddit-saas] $1k mrr in 6 weeks after years of nobody noticing the stuff i made — https://www.reddit.com/r/SaaS/comments/1ui80ik/1k_mrr_in_6_weeks_after_years_of_nobody_noticing/ — $1k MRR anecdote without enough hard detail to beat stronger items. - [R] [reddit-saas] founder here: shipped my devtool saas MVP, but I have 0 distribution. how do you guys get the first 10 users to validate it? — https://www.reddit.com/r/SaaS/comments/1uilkxz/founder_here_shipped_my_devtool_saas_mvp_but_i/ — Devtool distribution question, interesting but still mostly request-for-advice noise. - [R] [reddit-saas] My first paying customer — https://www.reddit.com/r/SaaS/comments/1uieekv/my_first_paying_customer/ — First customer milestone without broader lesson or evidence. - [R] [reddit-saas] Launching a mobile B2C SaaS (SnakeStreak) via pre-registration. Giving away Lifetime Pro to fight day-one churn. Thoughts? — https://www.reddit.com/r/SaaS/comments/1uijp8q/launching_a_mobile_b2c_saas_snakestreak_via/ — Pre-registration launch tactics post, but too speculative. - [R] [reddit-saas] Should I triple my sales volume with a $9.99 tier if it kills my customer satisfaction? — https://www.reddit.com/r/SaaS/comments/1uiehal/should_i_triple_my_sales_volume_with_a_999_tier/ — Pricing thought experiment without evidence beyond opinion. - [R] [reddit-saas] whats your process for checking ai-built apps before you ship to a client? — https://www.reddit.com/r/SaaS/comments/1uiimwe/whats_your_process_for_checking_aibuilt_apps/ — Question about QA for AI-built client apps, but the linked Reddit page was blocked during browsing and the item lacks enough signal on its own. - [R] [reddit-saas] Can I just edit my own launch video or do i actually need a launch video company ? — https://www.reddit.com/r/SaaS/comments/1uim25h/can_i_just_edit_my_own_launch_video_or_do_i/ — Video production advice request, not a memo item. - [R] [reddit-saas] How would you run a closed beta for a GPU sharing platform? — https://www.reddit.com/r/SaaS/comments/1uiluv6/how_would_you_run_a_closed_beta_for_a_gpu_sharing/ — Closed-beta discussion prompt, still too early-stage and generic. - [R] [reddit-saas] Stop building your MVP like it's a Series B product — https://www.reddit.com/r/SaaS/comments/1uhz0n5/stop_building_your_mvp_like_its_a_series_b_product/ — Generic MVP advice post, not new enough to keep. - [R] [reddit-saas] Uff... I'm tired of explaining my work to AI over and over again. — https://www.reddit.com/r/SaaS/comments/1uil080/uff_im_tired_of_explaining_my_work_to_ai_over_and/ — Complaint about repeating AI context, but no concrete method or product change. - [R] [reddit-saas] First Saas (Openpte.com) - Beta version — https://www.reddit.com/r/SaaS/comments/1uikx86/first_saas_openptecom_beta_version/ — Beta launch self-promo, too thin. - [R] [reddit-saas] Anyone know any good affiliate platforms specifically for SaaS applications? — https://www.reddit.com/r/SaaS/comments/1uikrta/anyone_know_any_good_affiliate_platforms/ — Affiliate platform question, not evidence. - [R] [reddit-saas] I’m looking for someone highly skilled in video editing, especially for SaaS / software explainer videos. — https://www.reddit.com/r/SaaS/comments/1uikpxq/im_looking_for_someone_highly_skilled_in_video/ — Hiring request for video editing help, not memo material. - [R] [reddit-saas] Why do we still waste 1 hour waiting at salons? Today I went to a salon and ended up waiting almost an hour because several people were ahead of me. I found apps like Salonist, Fresha, and Booksy, but they seem to focus more on salon management (appointments, billing, CRM, inventory, etc.) than sol — https://www.reddit.com/r/SaaS/comments/1uikns4/why_do_we_still_waste_1_hour_waiting_at_salons/ — Salon wait-time idea pitch, still at problem-statement stage. - [R] [reddit-saas] Traveling with kids just got much more fun with SAGA, native development question — https://www.reddit.com/r/SaaS/comments/1uikipt/traveling_with_kids_just_got_much_more_fun_with/ — Travel app promo and dev question, outside scope. - [P] [google-ai-changelog] Interactions API — https://ai.google.dev/gemini-api/docs/interactions-overview?hl=vi — Google says the Interactions API is now the default path for new Gemini apps, with server-side state, background execution, and better cache economics. - [P] [hn-top] HackerRank open sourced its ATS. My resume scored 90/100. Oh wait 74. No – 88 — https://danunparsed.com/p/hackerrank-open-source-ats — A reproducible writeup shows HackerRank's open ATS can swing resume scores widely across identical runs, which makes LLM-only screening look too noisy for hard cutoffs. - [P] [hn-top] GLM 5.2 beats Claude in our benchmarks — https://semgrep.dev/blog/2026/we-have-mythos-at-home-glm-52-beats-claude-in-our-cyber-benchmarks/ — Semgrep reports GLM 5.2 beat Claude Code on its IDOR benchmark under a minimal prompt harness, but the claim is vendor-authored and limited to one task. - [R] [hn-top] Age verification is just a precursor to automated attribution of speech — https://nonogra.ph/age-verification-is-just-a-precursor-to-attribution-of-speech-06-29-2026 — Opinion essay on age verification, but the EFF piece carried the stronger practical signal. - [R] [hn-top] Historical memory prices 1960-2026 — https://dam.stanford.edu/memory-prices.html — Memory price history is interesting reference data, not tomorrow-morning action. - [R] [hn-top] 5k menus from the New York Public Library’s Buttolph Collection (1880-1920) — https://pudding.cool/2026/06/menu-story/ — Historical menu collection project, off-topic for this memo. - [R] [hn-top] I used Claude Code to get a second opinion on my MRI — https://antoine.fi/mri-analysis-using-claude-code-opus — Medical self-use story with risky framing, not something to amplify. - [R] [hn-top] Knowledge Distillation of Black-Box Large Language Models (2024) — https://arxiv.org/abs/2401.07013 — Older black-box LLM distillation paper resurfaced on HN, but not a fresh development. - [R] [hn-top] Deciphering basmala — https://blog.plover.com/lang/bismillah.html — Language-history post, outside scope. - [R] [hn-top] Show HN: Zanagrams — https://zanagrams.com/ — Duplicate HN entry for Zanagrams. - [R] [hn-top] Tokenmaxxing is dead, long live tokenmaxxing — https://12gramsofcarbon.com/p/agentics-tech-things-tokenmaxxing — Essay on token economics and agent tooling, but not concrete enough versus stronger workflow items. - [R] [hn-top] The Baffling World of Masayoshi Son's Presentations (2020) — https://www.bloomberg.com/news/features/2020-06-23/golden-geese-and-unicorns-inside-the-eccentric-presentations-of-masayoshi-son — Old Bloomberg feature recirculating, not actionable. - [M] [hn-top] The KIDS Act would require age checks to get online — https://www.eff.org/deeplinks/2026/06/kids-act-would-require-age-checks-get-online — EFF argues the proposed KIDS Act would pressure platforms into broad age checks and more restrictive moderation, especially around private messaging. - [P] [hn-top] Model Training as Code — https://aleph-alpha.com/en/blog/model-training-as-code/ — Aleph Alpha describes an internal “model training as code” stack that treats training pipelines like CI-driven software, with immutable artifacts and nightly end-to-end validation. - [R] [hn-top] Professor denounces mass AI fraud on an exam at Brown — https://english.elpais.com/education/2026-06-28/ai-fraud-at-brown-university-academic-integrity-is-at-risk.html — AI cheating in higher ed is real, but this item does not change builder decisions here.