Peter Levels on Indie Startups, AI, and Building Fast

Levels explains why he builds and launches small products on his own, without venture funding, and tests demand through paying customers. The conversation moves through creative work, exercise as a source of purpose, his preference for building over managing, and his views on scaling.

Levels recounts feeling isolated and depressed after university and traveling, and how his father’s advice to do physical work influenced him to start building. He explains the 12-startup challenge, his first project Play My Inbox, and how tight time limits shaped product scope, payment and promotion.

Levels discusses feeling like a loser while traveling alone, the psychological cost of being detached from home and community, and the appeal and limits of digital nomad freedom. He recalls working through the night with friends in Bali, talks about Thailand and other travel experiences, and describes how Nomad List began as a way to connect with other nomads.

The conversation follows the idea stage in Levels’s startup process. He recommends noticing problems in daily life, learning from solutions encountered while traveling, and keeping a list of ideas. He distinguishes problem-led ideas from experimenting with new technology, which can amount to a solution looking for a problem.

Levels demonstrates Photo AI and describes its image generation, use of Replicate, uneven results and safeguards against inappropriate images. The discussion expands into his use of HTML, jQuery, PHP and SQLite, his reasons for avoiding unnecessary complexity, and his skepticism about claims that frameworks are always better.

Levels traces his AI image projects from Stable Diffusion experiments and This House Does Not Exist to Interior AI, AI avatars and Photo AI. He discusses fine-tuning, early demand and manual fulfillment, competition from larger apps, training data, image quality, and using customer choices to test model settings. They also discuss open-source models and AI development platforms.

Levels explains that he often chooses ideas based on what he can build, and shares speculative interests such as robotics and asteroid mining. The conversation considers technologies that have not yet found clear uses, subconscious idea generation, timing, and the value of staying open rather than trying to control every life decision.

Levels describes starting Nomad List as an editable Google spreadsheet and turning it into a website. He advises using whatever tools are available—including spreadsheets and website builders—then validating demand before worrying about the technology stack. The discussion also considers how crowdsourcing can organize contributions.

Levels explains Hoodmaps, where users color areas of cities by category and add or vote on local descriptions. He describes its crowdsourced data, how he organized map contributions, and its viral use. He says map hosting has been expensive and monetization difficult, while emphasizing that not every project needs to make money.

Levels explains that Nomad List began by comparing cities for remote workers, including internet speed and cost of living. He discusses its mixture of public datasets and community information, member profiles and meetups. The conversation turns to the practical problems of finding reliable accommodation, internet and workspaces while traveling.

Fridman and Levels discuss poor airline and hotel booking experiences, possible effects of bureaucracy, incentives and management, and how companies test changes. They consider whether internal competition between teams and greater freedom for engineers could help established organizations ship improvements.

Levels describes trying to learn virtual reality and 3D by building projects, and argues that practical action is more effective than waiting to master a subject first. They discuss using AI as a learning aid, self-directed coding challenges, persistence through frustrating problems, and focusing on the next step.

Levels says criticism can reveal specific product problems if it is separated from the emotion of being attacked, while noting the need for a thick skin and tools such as muting. He discusses Hoodmaps reaching Reddit and the role of TikTok, short-form video and social platforms in bringing users to products without paid advertising.

Levels advocates asking users to pay early, setting prices that can support an independent business, keeping costs low, and negotiating with service providers. The conversation moves to free-user abuse, 4chan’s influence and risks, and ways to limit harmful contributions while preserving open participation.

Levels describes using GPT-4 to assess city reviews and moderate a large chat community, including interpreting context and humor. They discuss AI-assisted fact-checking and the possibility of ranking social media replies by qualities such as argument and relevance, rather than relying only on chronology.

Levels explains his preference for automating recurring work rather than hiring a team, using Nomad List meetups as an example. He describes cron jobs, health-check pages, uptime monitoring and alerts sent to Telegram, and discusses extending automation to tasks such as podcast publishing.

Fridman and Levels discuss the effects of social media on attention and mood, including the way short videos can leave users feeling distracted. Levels describes muting political content and using browser extensions to hide views, recommendations and other distracting elements.

Levels discusses acquisition negotiations, changing valuations and why automated, profitable products may not be worth selling at a low multiple. He has a particular attachment to Nomad List and its mission, and reflects on the meaning of continuing to work rather than retiring.

Levels explains why he usually works alone, while describing a current project with his friend Andre and a past collaboration on Photo AI. The discussion explores how developers can adapt to an existing codebase and style before proposing changes, and turns to the design of Levels’s interfaces.

The conversation examines Levels’s iterative interface design and his practice of deploying small changes directly to production with checks and linting. They contrast this fast feedback loop with larger organizations’ slower approval processes, using YouTube’s multilingual audio feature as an example of a potentially important product change.

Levels discusses VS Code, Sublime Text and why he considers familiarity more important than choosing a fashionable editor. He and Fridman consider when to explore frameworks, how to learn their concepts through projects, and the difficulty of identifying a standard approach in a fast-changing software ecosystem.

Levels recounts switching from Windows to Mac before traveling and compares the systems, including the command line and development environment. He describes working on a couch with one laptop screen after dealing with repetitive strain, and argues that more equipment does not necessarily mean greater productivity.

Levels describes his usual schedule and the more intense stretches he uses to create new projects, including coding late, working alongside a friend, listening to techno and drinking coffee. The conversation covers concentration, silence, caffeine, naps and the importance of learning how one’s own body responds rather than following generic advice.

Levels describes Remote OK, which grew out of a jobs page on Nomad List. Companies pay to post remote jobs, and he explains how job-post upsells performed during a period of stronger demand before revenue fell and later rose again.

Levels recounts selling and giving away possessions before traveling, including the emotional significance of selling a camera tied to music projects with friends. He reflects on the freedom and trade-offs of owning little, travel and university choices, and the relative value of relationships, experiences, exercise and meaningful work.

Levels explains why he closed direct messages, while acknowledging the meaningful stories and people he cannot meet. He discusses helping people who take initiative, learning through action rather than endlessly seeking advice, managing time, and how friendships can persist despite infrequent contact.

Levels says he prefers people to share their genuine interests rather than confine themselves to one public topic. The discussion of a post about NVIDIA’s origins leads to cafes as places for work, conversation and chance encounters, and to differing attitudes toward entrepreneurship in Europe.

Looking at his project history, Levels discusses differences he sees between European and US companies, regulation and barriers to starting an AI business in Europe. He supports a technology-forward outlook through EOC and EUOC, arguing for more building and experimentation while acknowledging that societies also face problems.

Asked for advice, Levels encourages young people to try their own path, accept that it may not work immediately, and commit to what they genuinely want to build. Fridman thanks him, then closes the episode with a line attributed to Dropbox co-founder Drew Houston.