Your customers are changing slower than the models
"It is neither wealth nor splendor, but tranquility and occupation which give you happiness."
― Thomas Jefferson
* Today’s newsletter is brought to you by Hashiri.AI.
If you’ve ever pasted splits into ChatGPT and typed “how was this run,” you know the routine: explain everything, every time. Hashiri.AI already knows. It syncs Garmin or Apple Health, scores your readiness each morning from sleep and HRV, and plans today’s run to match, with your race in sight. You just ask.
Free to start at hashiri.ai.
Hi friends,
We hand-picked 3 good pieces of content to inspire and motivate you now and in the future. Keep track of the insights that resonate with you by highlighting them with Glasp💡
If you want to reread or highlight this newsletter, save it to Glasp.
📚 3 Good Recommendations
Agent swarms and the new model economics
by Wilson Lin (17 mins)
Coordination is the real cost: On the same task, the old swarm sprawled to 54 crates and generated 70,000 merge conflicts before it had to be paused. The redesigned one settled on nine crates early and logged under 1,000 in four hours.
Not every step needs a frontier model: Opus 4.8 planning with Composer 2.5 doing the work came to $1,339. GPT-5.5 in both roles came to $10,565. Very few moments in a large task genuinely require the most expensive intelligence.
Less code, same result: Both swarms passed the same test suite for a SQLite implementation. The old one needed 64,305 lines of engine code. The new one needed 9,908.
How to Apply Tiny Experiments: A Field Guide to Curiosity-Driven Learning
by Glasp (12 mins)
A PACT replaces the goal: Instead of “get 1,000 subscribers,” you commit to “I will publish one short essay every week for eight weeks.” Purposeful, Actionable, Continuous, Trackable, and entirely inside your control.
Outcomes become data, not verdicts: A missed goal arrives labeled as failure. A finished experiment just produces findings, and that reframe lowers the stakes enough that you actually start the next one.
Plus Minus Next: What went well, what drained you, what you will try next. Run weekly, this is what turns a string of experiments into a direction instead of a random walk.
This is AGI: Sequoia AI Ascent 2026 Keynote
by Sequoia Capital (32 mins)
A revolution in computation, not communication: The internet, the cloud, and mobile were all about how information gets distributed. AI is the first wave about how information gets processed, and that is why it behaves differently from every platform shift before it.
Build toward your customers, not toward the capabilities: The counterintuitive part of their MAD framework (Moats, Affordance, Diffusion). New model capabilities arrive constantly and what you build on them may be irrelevant tomorrow. Your customers change far more slowly, so wrapping yourself around them is the more durable moat.
The diffusion gap is the opportunity: Capabilities spread into the market far slower than they are created, and every day the frontier models move faster than the average Fortune 500, that gap widens. That gap is where application-layer companies get built.
📣 Community Updates
🟦 75,000 Books and 80,000 Posts:
The community has now brought over 75,000 Kindle books into Glasp and published more than 80,000 Posts. Highlights are where reading starts, and Posts are where it turns into something worth sharing. Import your books here.
🟨 13 Ways In, 8 Ways Out:
Glasp now imports from 13 sources, including Kindle, Pocket, Matter, Wallabag, Instapaper, and plain CSV, and exports to 8, including Notion, Obsidian, Roam, Tana, Workflowy, and Zotero. Your highlights should not be trapped inside any one tool, ours included. Set yours up here.
🟥 Turn Your Highlights Into Flashcards:
Anki Export is new. Send a passage you highlighted straight into Anki and it becomes a spaced repetition card, so the thing you thought was worth marking is still with you in six months. Try it here.
🟩 Automate Glasp With Zapier:
Glasp on Zapier now covers four triggers and two actions: New Bookmark, New Document, New Highlight, New Summary, plus Create Bookmark and Create Highlight. Log every highlight to a spreadsheet, file summaries into Notion as they appear, or push links in from wherever you find them, with no code. Browse the workflows here. If you would rather build directly, the API docs are here.
🤝 Ask
Two small favors, and both take about a minute. Reviews are how new readers decide whether to trust us, and rating counts weigh heavily in how these stores rank and recommend.
🟥 Review the Chrome Extension:
Hubros wrote this in June without us asking for it. Reviews like this are the main reason someone new decides to try Glasp at all, and they take about a minute. Leave yours here.
🟨 Rate the Mobile App:
The iOS and Android apps are far newer than the extension, so each rating counts for much more there. Rate it on iOS here and on Android here.
❤️ Gratitude
Thank you for sharing and mentioning us on X, LinkedIn, and/or your blogs. 🙂 We appreciate your support! Please don’t hesitate to ask us anything at any time. Also, feel free to join our Reddit Community ;)
Glasp AI Writing: AI-Powered Research & Content Creation Tool on AiToza
Glasp on Your Trusted AI
YouTube Summary & ChatGPT by Glasp on Chrome-Stats
Glasp Web Highlighter: PDF & Web Highlight on EXTPOSE
Glasp on Aona AI
My browser was destroying my focus — these AI extensions fixed it on inkl
생산성 개선 (Productivity improvement) on GitHub (kodingwarrior)
YouTube Summary & ChatGPT by Glasp 2.1.1 for Google Chrome 擴充功能 on 軟體王 Softking
We hope you enjoyed reading this newsletter!
See you next week ;)
Best,
Kei and Kazuki
--
Would you like to take Glasp on the go?
With the Glasp mobile app, you can highlight and organize your favorite content anytime, anywhere. Stay productive on the move and never miss an insightful quote.
Partner With Glasp
We currently offer newsletter sponsorships. If you have a product, event, or service you’d like to share with our community of learning enthusiasts, sponsor an edition of our newsletter to reach engaged readers.
The number that stuck with me from the Cursor piece: same task, same passing test suite, $1,339 with Opus 4.8 planning and Composer 2.5 doing the work, against $10,565 with GPT-5.5 doing both roles. Nearly 8x, for the same result. Very few moments in a large task actually need the most expensive model, and most of us are still paying as though they all do.
The Sequoia keynote arrives at a similar place from the other direction. Capabilities move weekly, your customers do not. Wrapping yourself around the customer is the slower, duller, more durable bet.
And an easy one to answer: of the 8 places you can now export highlights to (Notion, Obsidian, Readwise, Zotero, Tana, Workflowy, Roam, Anki), which one do you actually live in? It genuinely decides where the next round of work goes.
No Substack account? We are also on Reddit: https://www.reddit.com/r/glasp/