TypingMind Ships Local Code Execution as Founders Debate Whether AI Kills SaaS

Key Takeaways

1. TypingMind released what it calls its strongest update yet, running code locally in the browser and handling PDF, docx, xlsx, image, and video files. 1
2. A wave of founders and operators argue AI subscriptions are replacing stacks of SaaS tools, with some canceling nearly all business and personal subscriptions. 1 2
3. The "software is dead" debate centers on personal software: AI agents building tools on demand, while skeptics point to maintenance burdens and ordinary users' limits. 1 2 3
4. Noah Kagan says he replaced an $11,000-per-year Carta bill with a self-built, open-source tool finished overnight. 1
5. Marketplace listings show small, automated internet businesses changing hands, including a US lottery app at a $152,498 asking price and a UK poker YouTube channel at $124,740. 1 2
6. Alex Lieberman says AI Agent Sprints drew hundreds of enterprise inquiries, including Tesla, Airbus, Amazon, and Samsung, while many companies stall when moving agents into workflows. 1
7. Founders shared AI-driven growth stories, including a five-salon owner whose AI agent tripled late-night leads and an 18-year-old with no coding experience claiming $50,000 monthly revenue. 1 2

1. AI Products and the SaaS Displacement Debate

1. TypingMind shipped an update it describes as its most powerful, adding local in-browser code execution and support for PDF, docx, xlsx, images, and video; the same author says building with AI is fun and he likes polishing his own products to fine detail. 1 He also argues a personalized approach beats bloated SaaS: a typical product has 50 features while he needs about six, two of which are missing or locked behind enterprise tiers, and vibe coding a replacement takes roughly a day. 2
2. @levelsio frames AI subscriptions as substitutes for multiple SaaS bills, saying people would rather pay $20 a month to Anthropic or ChatGPT than pay separately for Adobe, Canva, Microsoft 365, Grammarly, and Squarespace. 1 He says he has canceled nearly all enterprise SaaS subscriptions, keeping only AI inference, Hetzner servers, Cloudflare storage, Backblaze backup, and taxes, and personally keeping only Claude Code, YouTube, Spotify, Starlink, an eSIM, personal cloud storage, utilities, and taxes, predicting a "cancellation wave" once people see the savings. 2
3. The "software is dead" argument, as he describes it, is about the personal software era: AI agents will build software on demand to solve a user's problem rather than users asking for software, creating a market where agents pay to connect to software and data via MCP or APIs but push prices to the lowest or free tier. 1 He says @Grok summarized his view as "directionally right but slightly early," agrees, and expects years before global adoption of AI chat apps and agents. 2 He also responds to critics by translating AI-company claims that "now is the best time to start a company" into "now is the best time to pay for tokens to start a company," noting roughly half his net worth is in AI stocks including Google, SpaceX, Nvidia, and tech ETFs, while arguing this is the best era for creativity, not necessarily for software revenue. 3
4. He cites a scale gap: about 40 million software developers worldwide versus roughly 2 billion people regularly using AI tools, a 50x difference that could reach 6 billion people and 200x, with AI agents already able to do development work and likely to lower all costs while developers may not be paid for the same work. 1 He expands his "remaining moats" list to proprietary data or content, network effects or community, audience or distribution, physical components, capital especially in AI companies, and regulatory licenses or legal barriers, while saying technology moats have largely disappeared and pointing to @omarchy as an example of personal software that is malleable, user-controlled, free, open source, and sponsor-supported. 2 3 4
5. Counterpoints in the same thread argue personal software is not yet mainstream because non-technical people lack time to maintain it and most "personal software" remains one-off, though agents and tools may advance until apps maintain themselves. 1 Others argue ordinary people's AI ability is overestimated, that many struggle even to set up an app and will pay someone to handle it, and that services like finding flights still require human help, so plugins and MCP servers remain software businesses. 2 Additional replies note models still make many mistakes, large ERP systems with business rules, legal, and financial calculations have high error costs, and social apps are safe because users buy the network itself and community building is a moat. 3 4 5

2. Builders Replacing Paid Tools and Sharing AI Growth Stories

1. Noah Kagan says he replaced Carta, which cost $11,000 per year, with a self-built tool called SumoShares completed in one night, supporting cap tables, option grants, exercises, 409A reminders, audit logs, and CSV export, and released it as open source and free. 1 Separately, Tony Dinh argues for personalization over feature bloat, saying a typical SaaS has 50 features while he needs about six, two of which are missing or enterprise-locked, and that he lacks time to hunt alternatives and does not want data locked into a SaaS vendor when vibe coding a tool takes about a day. 2
2. Codie Sanchez shares a case in which Karen, who owns five salon locations, took her AI course with her son and used an AI agent on the website and social channels to capture late-night inquiry leads, tripling lead volume, driving revenue growth, and benefiting all locations. 1 On Starter Story, 18-year-old Lino with no coding experience claims $50,000 monthly revenue from a fitness app and advises founders not to spend on small influencer partnerships but to trade equity for a partnership with one major fitness influencer to help inexperienced founders start. 2 A separate 19-year-old, Vadim, claims $11,000 monthly revenue from a SaaS, saying he worked at a top local marketing startup in high school at 17, quit, built a web app with no coding experience, and forced himself to start by "burning all bridges." 3
3. A reposted account says that eight months ago at 18 the author encountered vibe-coding, quit after being inspired by Starter Story and "burned the bridges," got a first sale six months later, then cut UGC marketing and reached over $10,000 monthly revenue, passing 90,000 users yesterday, with no technical or programming experience. 1 The account also reposted that a related video had been published. 2

3. Enterprise AI Adoption and Marketplace Signals

1. Alex Lieberman says AI Agent Sprints gained traction within weeks, drawing hundreds of enterprise inquiries including Tesla, Airbus, SharkNinja, Amazon, Palo Alto Networks, and Samsung, with a few spots left in the October cohort. 1 The five-day free bootcamp for enterprise leaders has three phases, finding opportunities, mapping workflows, and building prototypes with customer data, producing an AI roadmap with ROI, up to eight mapped opportunities, and up to four working prototypes. 2 It targets VP-level and above decision-makers at companies with over $100 million in revenue or funding, is free, and is led on-site by FDEs and strategists, with the account claiming a repeatable process can bring agents into production in about two weeks. 3 The same account says many companies have deployed LLMs company-wide, done AI enablement, and used coding agents widely in engineering, but most stall at the next step of bringing agents into organizational workflows. 4
2. Marketplace listings show small automated businesses for sale: a US iOS lottery app operating about a year with high automation, users scanning tickets and checking results, about $50 monthly hosting cost, $68,200 annual revenue, $3,300 monthly profit, 1,017 paid subscriptions, and a $152,498 asking price. 1 A UK poker entertainment YouTube channel launched in 2024 with a lean low-overhead model, 202 published items, $64,700 annual revenue, 27,300 subscribers, $23,500 annual profit, and a $124,740 asking price in First Access. 2 Other listings include a UK trading education and software business operating since 2022 with an FCA advertising account, $2.78 million annual revenue, $76,000 monthly profit, 106 active subscriptions, and a $4,999,051 asking price, and a Nordic brand operating about 3.5 years selling premium sneaker display boxes through online stores, marketplaces, and trade shows, with $341,000 annual revenue, $5,200 monthly profit, 25% repeat buyers, and a $185,000 asking price in First Access. 3 4 A US faith-based YouTube channel operating two years with an editor handling production and the owner off-camera has $78,000 annual revenue, 584,000 subscribers, 341 million total views, and a $150,000 asking price in First Access. 5
3. On acquire.com, listings include a Shopify upsell app with $255,000 ARR, 143% year-over-year growth, 97% gross margin, $187,700 TTM revenue, and a $950,000 asking price; an AI staffing SaaS for UK rapid hiring and driver dispatch with $36,000 TTM revenue and a $98,000 asking price; an AI bank statement extraction SaaS for accountants doing balance checks in Google Sheets and Excel with $52,400 TTM revenue and a $225,000 asking price; and an AI stock research and alternative data platform SaaS with $212,000 ARR, $181,000 profit, $269,500 TTM revenue, and a $949,000 asking price. 1 2 3 4 The account also reposted a view that a bootstrapped company reaching $1-2 million in revenue, staying profitable, and selling for $5 million with the founder owning 100% can net more than a $20 million sale where the founder keeps only 15% equity, and said selling a business is complex enough without platforms adding difficulty, advocating clear processes, fair treatment, and no games. 5 6

4. Moats, Markets, and Founder Commentary

1. Alex Lieberman says technology moats are basically dead and adds two moats, brand as trusted distribution and regulatory positioning, while the quoted post says proprietary data, network effects, and physical components are the only remaining moats. 1 @levelsio similarly lists proprietary data or content, network effects or community, audience or distribution, physical components, capital especially in AI companies, and regulatory licenses or legal barriers, while acknowledging regulation is a moat but noting ChatGPT, Anthropic, and Google are entering it and that holding Google stock is at least a way to participate. 2 3
2. Sam Parr asks newsletter operators whether Substack or other newsletter platforms bring substantial growth through their networks, saying Substack's main advantage seems to be subscriber growth via its network and asking whether that matches others' experience. 1 He also comments on a YouTube creator announcing retirement to spend time with children, play Warhammer and Lego, and write books, calling it courageous to say "I've had enough, even if income drops tenfold, I want to do what I love," and predicting such cases will multiply a thousandfold as AI takes over, from which he infers a huge opportunity in making toys for adults and even considers moving from B2B SaaS in that direction. 2
3. Nick Huber advises first-time founders not to aim for "scale" or changing the world but to start small, choose low-risk companies, practice through low-risk decisions, allow mistakes and survival, and accumulate operating skills, capital, and networks for later bigger opportunities. 1 He also reposted a message about hiring globally to build a better team at lower cost and directing viewers to Somewhere. 2 Justin Welsh reposted a view that the richest people he knows drive ordinary cars and live in ordinary houses but have unusually large amounts of discretionary time, and argues the opposite path to success is simple: stay optimistic, be kind to everyone, do not let politics dominate you, do not follow outrage peddlers, always do excellent work, eat and sleep well, and train hard, while calling the biggest power move a slower life while still getting rich. 3 4 5
4. Patrick McKenzie uses a dialogue to argue that in technical troubleshooting, email replies should carry near-zero weight and one should first confirm whether the other party is communicating with a person or only reading documentation; when asked why he assumes they would lie, he gives no further response. 1 Arvid Kahl contrasts Turing's 1950 prediction that machines would enter every field of human intelligence and eventually compete as equals with a 2026 joke about mathematicians and OpenAI, and separately argues software is not dead because enterprise customers seek multi-year supply contracts and integrations, while personal software will struggle for a long time to fit process-driven hierarchical enterprises. 2 3 He also reacts to news that hackers remotely defrosted hundreds of French supermarket refrigerators and caused massive food spoilage with "Not cool!" 4 Tony Dinh says social media has been ruined by AI comments. 5 Alex Lieberman reposted a recruiting message from @tenex_labs seeking nonlinear thinkers, builders, and engineers to submit the hardest problems they want solved under fully autonomous conditions, saying tenex_labs is not a consultancy but connects people with ethical AI and helps organizations transform, and is hiring. 6