Key Takeaways
- A forwarded argument says most teams have not built a real "software factory"; they have only made AI-assisted shipping easier, and the workflow is defined by one shared pipeline rather than by output.
- BizScout reports 183K+ users, 70K buyer-seller connections, 1.5M listing views, a $5M seed round, and a Transworld partnership, plus a score showing over half of listed car washes cannot service an SBA loan at asking price.
- Noah Kagan says Claude Max at $200/month replaced $11,000/year Carta and $4,788/year Zapier, cutting $15,788 in annual software spend to $2,400.
- Two app founders describe prompt-built MVPs and paid-social scaling: AbMaxx at about $55K MRR and Canary at 50K downloads with 2,000 paying subscribers.
- Acquire.com lists small startups for sale and argues that cutting $100K of wasted spend can add $400K of value at a 4x profit multiple.
- Levelsio reports an iOS 27 filter regression and says slow game loading is a main reason he ports games to the web.
1. AI-Assisted Software Factories
- A forwarded post argues that most people have not actually built a software factory system; they have only made AI-assisted publishing easier, and the category is misjudged as nonexistent because of a lack of vision. The factory is defined as all work passing through the same process stages, standardized by process rather than output, with the pipeline starting from a "signal" such as a Slack discussion, Linear ticket, Sentry alert, CI failure, tweet, or idea. Signals enter either through a person who picks them up and builds with Droid, or through a workflow that triages by a set goal, has a droid write a spec, and opens a PR; most work is still human-initiated, but the workflow share is growing. Every change, human- or AI-written, passes the same checks: Droid review, security review, an author-blind "risk judge," CI, and an agent doing QA in a live preview, with the goal of raising the "no-regrets merge rate" until most changes can merge automatically. If something breaks after launch, the responsible PR is traced and fed back as a new signal, which also tests whether the checks worked, for example when the risk judge approved a change that later failed. The post argues for an autonomous-driving-style release logic: automatic merging is allowed only when AI-written changes cause fewer failures than human changes and this is measurable; most changes still keep human approval today, but automatic merging, workflows, and risk scoring together let the system improve itself from the signals it tracks, with humans setting rules and supervising 1.
2. Small Business, Acquisitions, and Software Spend
- Codie Sanchez describes BizScout's two-year progress: more than 183K users, 70K buyer-seller connections, 1.5M listing views, a $5M seed round with participation from Valor, Balaji Srinivasan, Sean Rad, and Bill Perkins, a partnership with Transworld that put all of its active listings on the platform, and the launch of a 0-100 BizScout Score, a Private Client service, and AI off-market deal alerts. She says the BizScout Score shows that more than half of listed car washes cannot service an SBA loan at asking price, and that 203+ Contrarian Academy members completed business acquisitions this year, 47% of them while still holding full-time jobs at closing. Member examples include Erin, who after cancer treatment quit her job and bought a 20-year-old family laundromat for $65K after following the original owner for over a year, later finding employee cash theft and actual revenue of $250K rather than $220K, and another member who helped Tim avoid a $1.4M acquisition. Her team was also acquiring businesses during the same period, encountering fraud that killed a deal, two acquisitions that nearly closed, and a seller who withdrew at the last minute, which she uses to argue that buying a business or starting one is slow and messy and often involves people who waste your time 1. She also argues that most "you can't do it" advice is just failures defending themselves, and that people should not overthink the starting direction but act quickly and keep asking whether their life is what they want 2 3.
- Noah Kagan says Claude Max at $200/month has replaced his $11,000/year Carta and $4,788/year Zapier; of Zapier's 123 automations, 2 did 88% of the work and were rebuilt with Claude. He frames this as $15,788 in annual software spend replaced by $2,400, before counting productivity gains, and concludes Anthropic will inevitably raise prices 1.
- Acquire.com lists small startups for sale: a migraine prediction and management app at $307K TTM revenue asking $380K; a multi-model AI chat app at $20.3K TTM revenue asking $57.3K; a creator-focused AI inbox SaaS at $192K ARR, $12K monthly profit, $42.2K TTM revenue, asking $125K; and a no-code AI app and agent platform for creators at $51.4K TTM revenue asking $72K 1 2 3 4. A forwarded post argues that a SaaS with $2M ARR and $500K profit that cuts $100K of wasted spend reaches $600K profit, and at a 4x annual profit multiple its value rises from $2M to $2.4M, adding $400K without new customers, while acknowledging that businesses differ and buyers consider more than profit 5. Acquire.com argues founders often underestimate how margin improvement lifts sale valuation and that stopping wasted spend can sometimes make a company more valuable more easily than growing revenue 6. It also argues that what buyers pay in 2028 depends heavily on the business's state in 2026-2027, with clean books, lower founder dependence, better growth, and fewer surprises all accumulating before a listing 7.
3. App Launches, Growth Tactics, and Platform Bugs
- Starter Story profiles two app founders. Lino Leiden, 18, dropped out of college and started building apps in February; he spent two months building V1 in Lovable through prompts, was unhappy with the quality, rebuilt V2 in Rork in one month, and earned $7K in the first month after launch. AbMaxx lets users scan their abs, get body-fat percentage and a lower-ab score, and follow a customized plan; it is at about $55K MRR, 16K downloads in 28 days, and 3K+ paying active users, and the founder says there is no direct competitor. To go from $10K to $60K per month he stopped all UGC and influencer deals that returned only 5-10% profit, switched to an equity deal with one large fitness influencer for free advertising, increased Meta ads at 2-2.5x ROAS, and copied the creative to TikTok ads for 4-5x ROAS covering ages 15-30. His five-step restart playbook: join the X app-founder community; use SensorTower to find niches with multiple apps at $10K+/month; validate with faceless TikTok slideshows and a "comment APP" CTA; launch in Rork within 7 days; and pay micro-creators with 500-1K followers $100 for 10 videos to find a winning hook before scaling paid ads 1. Ben had never written code and launched his first app, Canary, 9 months ago; users learn languages through songs they like, train with AI, and practice with people, and the idea came from a video of him singing in French with translated lyrics that grew a page to 200K followers in under a year, after which he built a rough MVP in about a month. Canary has 50K downloads and 2,000 paying subscribers at $10/month or $79/year, with 60% choosing annual; its growth method is to post heavily, find winning content, make variants, boost with paid social, kill losers, and repeat 2.
- Levelsio reports an iOS 27 regression: with Zoom's Low Light filter and Smart Invert both on, the screen becomes brighter than with Smart Invert alone, white backgrounds turn light gray at about 70%, black text turns pure white, and the intended dark night-reading effect is lost. He suspects the two filters are swapped in the display compositing pipeline and includes reproduction steps, a fix task, and acceptance criteria, hoping an Apple employee will help fix it; the quoted post says the combination originally made the screen nearly as dark as a Kindle for night reading 1. He says a main reason for making web games and porting existing games to the web is that modern games load too slowly: his RTX 5090, 128GB DDR5, Ryzen 9 9950X3D, and 4TB NVMe setup still takes nearly 3 minutes to load Flight Simulator, after several minutes of installing updates, whereas games used to be playable within about 30 seconds or instantly; the quoted post says Battlefield 1942 has been upgraded from v1.0 to the 2004 v1.61 and is playable multiplayer in the browser with up to 64 players 2. He also says history is repeating: 9 years ago he built a site to mute keywords and import Twitter because social-justice and political content was so bad, and now the same problem is AI replies 3. He quote-posts an account of a German state-level fund rejecting an all-male three-founder team as "not diverse enough" and comments that the EU is like the US but still stuck in 2016 "woke" thinking 4. He says nickname reservations are now available across all games, emulators, and IRC at @pieter: a reservation is universal and cannot be taken by others, IRC unlocks it through NickServ with a code, files saved in Windows XP and 3.11 emulators persist, and an in-emulator mailbox may come later; the quoted post says he fixed the stretched portrait mobile interface of BF1942 by modifying the engine and relays Claude Code's account of HUD recognition, per-element measurement, and compressing height by 2.9x 5. He also shares a new web game where players shop in a supermarket and learn to avoid microplastics in containers, heavy metals in baby food, and forever chemicals in cookware, with research links on the options, voice support, and a baby-products section 6.
4. Work, Talent, and Career Signals
- Patrick McKenzie says he joined @TheZvi on the Complex Systems podcast to read Anthropic's S-1, the pre-IPO filing 1. Arvid Kahl argues that cloud agent work plus Starlink makes most development work doable anywhere and raises questions about mobility, pay, and always-on expectations 2. Nick Huber argues careers and startups are not linear: years of little progress can be followed by 6-12 months that deliver years of progress, and the key is staying in the game so the high-growth period can arrive; he says he has seen many people five years ahead because of one good year 3. He argues virtual assistants have long replaced lower-level work at lower cost, and that international talent plus AI will now replace higher-level work at far lower cost 4, and he worries that college graduates and ordinary white-collar workers are in a difficult position because overseas employees plus AI can outperform Americans at work 5 6. He recalls mentor Dan Cohen, who sold a foundation-repair company for nearly $10M, warning him during his first startup that if he did not like sales he should get a job, and relaying Cohen's view that every part of entrepreneurship is sales, including getting employees to trust and join, persuading suppliers to work with you, getting investors to trust you with money, and getting customers to buy, with the key being to understand what the other side needs and explain what you can provide 7. Justin Welsh argues that people ask entrepreneurs about a backup plan while nobody asks that about relying on a single job, reflecting a double standard between stable employment and entrepreneurial risk 8 9; that the longer you avoid the beginner posture, the longer you stay a beginner 10; that people who can do the work always need a job while people who can sell outcomes never need one, stressing sales over execution 11 12; and that the best things money brought him are time with his wife, better sleep, and the ability to clear his schedule completely 13.
