John Moberg

“$25 per PR review is crazy” - 750K TC

@HenrikTaro wow Word-changing

Claude Cowork showed the unlock: LLM + real computer. But legal workflows still stall on the last mile: DOCX editing. Most approaches are either (a) coarse libs (python-docx) → formatting/styles/numbering/tables drift, or (b) raw OOXML surgery → token-dense + brittle at 40+ pages. We built the missing piece. Thread 🧵

@ChrisLovejoy_ @readwise @obsdmd can probably bootstrap quite nicely with claude code-style dynamically generated questions!

@johnmoberg @readwise @obsdmd Yeah this feels like the holy grail for accelerated personalised learning. Probably possible today but with overhead of setting up your own personal system. Maybe soon there are tools that prevent you needing all that manual setup and config. Would be a beautiful thing haha.

One thing I'm looking to update is my 'content consumption' pipeline. Various iterations of this over time but until now basically: - save article to read-later (e.g., @readwise chrome extension / mobile app, @obsdmd web clipper) - read it and highlight interesting parts - auto saved as MD file inside @obsdmd vault for future reference Feels like this is too passive though and I want to bring LLMs into the process. Am considering something like: - save article to read-later (TBD where this is stored) - use skill to review and engage with these articles. summarize, read segments, discuss the ideas. - rather than 'highlights' focus is more on storing 'learnings' / 'takeaways' More active process, so should internalize the ideas better. And maybe use LLMs to re-surface and engage more with the ideas later too. x.com/ChrisLovejoy_/…

@ChrisLovejoy_ And then you can align an agent judge with the tastemaker and scale to infinity

The most efficient AI product eval approach is to have a single 'taste maker' who's intimately familiar with the prompts, outputs and failure modes. This approach can scale surprisingly far. I know several 50+ people companies doing multi million ARR with this approach.

opus 4.5 means ANYONE can now ship a feature-rich product like Jira in days instead of years. finally we have democratized suffering

Interesting way to frame people/projects. Alpha = upside, beta = volatility. For some projects you really want high alpha, but some need low beta (especially as they mature). Processes always decrease *both* alpha and beta, sometimes you really care about beta, but beware of alpha impact

@elitasson Do you have a ”make me feel weak and useless to fuel my workouts” personality for the coach?

6/ j4.coach My personal AI coach, built with React Native, interviews me and creates a custom workout schedule that aligns with my goals, gear, and lifestyle. It will be released on the iOS App Store in the next few days.

If you are so productive with Claude Code, where are all the new projects? Here they are! Here is what I've been spending my free time on this Christmas. All while having the flu and taking care of the family (2 kids under 4 years old).

@craigzLiszt at least we will have digital abundance

I fully expect all major powers to be at war by the end of 2026. I hope I am wrong.

@bcherny beautiful

run this: /mobile

cmon opus you're not on the clock, just ultrathink and come back when you're done

Codex plays classical. Opus plays blitz.

only 4% of doctors use claude code, we're still so early

@peer_rich 100%, it's also where people are used to really bad tech, incredible seeing their eyes light up when it actually works

tech infuses every industry  construction healthcare and farming all have elements of tech nowadays and even though some of these legacy industries are slow to adopt new things when they do its immediate my dads generation went from horses to tractors in only one generation its just like us buying the first iphone when the right tech arrives it can completely transform an industry youd be hard pressed to find a phone that doesnt build upon the foundation of an iphone traditional industries like farming healthcare and construction arent the typical tech targets but theyre where real impact happens these are places where tech can genuinely improve peoples lives not with another b2b saas tool or productivity app  the best problems to solve are the ones affecting billions of people daily and when you have massive struggling industries theyre ripe for innovation with enormous impact potential if you can succeed there

gpt-5.2-pro, but it keeps thinking for days and gives me a call when it wants some input

"The biggest improvement in my work environment since I started as a doctor (I’ve worked for 30 years)." great sign when the people usually resistant to change are the first to embrace it

@felixhhaas @ivanhzhao @antonosika 🚀

Designing products in the age of AI! Join @ivanhzhao, @antonosika, and myself in person to explore how AI is changing the way we design, build, and scale products. We’ll dive into: → The intersection of AI, design, and product → What the next generation of tools means → Stories of building products that empower millions Stockholm, Sweden 18:00 – 19:30, September 24th Comment if you want an invite. Seats are limited.

I want to frame this whole article.