Real lessons, monetization strategies, and new methods from people building and growing a one-person web or app business.
A WordPress plugin for real estate developers was built about 18 months ago. It lets visitors click apartment areas shown on a building photo, then see prices, availability, and floor plans. CodeCanyon rejected the plugin outright instead of asking for changes. The product then moved to a freemium model: a free core version on the WordPress plugin directory, with paid upgrades handled through Freemius. The first paying customer arrived within a week of launch. Total net revenue has now reached $9,830, which is not enough to replace a job but is real validation from a product one marketplace rejected.
Claude was used to choose a buildable app idea when there was no clear idea to start with. The chosen niche was skin progress tracking for people using acne treatments such as tretinoin or isotretinoin. The app lets people take selfies, then shows an AI score and a trend curve so they can see whether their skin is changing over time. The aim is to reduce guesswork when judging whether a treatment is working. The app is called Skincurve, was built in about two days, and has passed App Store review. Its privacy angle is that photos normally stay on the device, with data leaving only when analysis is run. The launch plan is to rely only on App Store search optimization and organic discovery, with no ads or social media push.
RecallSafe is a free iOS app for people who need food recall alerts that match their own allergies. Users choose their allergens and state once. The app reads official FDA and USDA recall data, filters it to the recalls that matter to that person, and sends a notification when a new match appears. It does not require an account, login, or tracking, and the user’s information stays on the phone. Each recall links back to the official government source. The core app is free and has no ads, with optional in-app purchases for extra features. An optional barcode scanner can help check products, but no match does not mean a product is safe because recall notices may not list every UPC.
Reading study notes, GitHub pages, documentation, research papers, or articles often leads to the same loop: copy a confusing part, open ChatGPT, paste or upload a screenshot, read the answer, then return to the original page. The workflow works, but switching tabs breaks focus. A new Chrome extension aims to remove that break by letting people highlight text on a webpage, ask AI inside the browser, and keep asking follow-up questions without leaving the page. It is still waiting for Chrome Web Store approval, but it is already functional and has been tested with a small group. Related small tools point to the same need: a clipboard manager that stores several copied items and pastes them in order, a meeting note overlay that stays on screen, an AI overlay for coding tests or online interviews, and a notes app that floats over other apps. The common idea is simple: keep the help where the work is happening.
HikeCampSeek is a website that helps people find reopened campsite and multi-day hike bookings in Australia and New Zealand. During busy seasons, popular spots can sell out quickly, and canceled places are easy to miss unless someone keeps checking booking pages again and again. The service watches supported booking systems and sends real-time notifications when a canceled spot or new opening appears. It also includes a trip planner for arranging routes and schedules across different regions. A new Great Walks map for New Zealand is in progress, showing trail routes together with campsite and hut locations so travelers can plan visually.
getbmicalculator.com is a free BMI calculator. It does more than show one BMI number after a person enters height and weight. It also shows the WHO risk category, such as underweight, normal, overweight, or obese. A visual scale shows where the result sits. The tool calculates a healthy weight range from the person’s height, using either pounds or kilograms. It accepts mixed units, such as height in feet and inches with weight in kilograms. It also explains why BMI has limits and should not be treated as a perfect health measure.
przm is a web service that tries to show a person’s investments as one complete picture. Its maker had money spread across 3 brokerage accounts, an employee stock plan, and an old 401k from 2 past jobs, so no single app showed the full picture. Each finance app only showed its own slice, and many holdings were hidden inside funds, which could make a supposedly balanced set of accounts depend heavily on the same few tech stocks. przm aims to scan the whole portfolio across accounts, show where money is too concentrated, and reveal overlap that one broker screen would miss. It also aims to show where money is moving across the market, not just whether prices are up or down, by highlighting sectors and smaller sub-sectors that money is entering or leaving. The goal is also to lower the barrier for people with little finance experience who find trading hard to understand.
Repeated desktop work often needs a written instruction file before an AI agent can do it later. Tasks like pulling reports, creating the same forms, and moving files are useful to automate, but writing every step, input, and check by hand takes time. This approach records the task once while it is done on screen, then turns that recording into a SKILL.md file an AI agent can reuse. It captures real screen actions and adds visual context from the interface. Values that change, such as search terms and dates, become reusable inputs instead of fixed text. The tool runs as an MCP server, so it can connect to Claude Desktop or another MCP client. The available actions include starting a recording, stopping it, compiling it, and listing saved items.
In early SaaS work, building the product can feel predictable. Customer acquisition is much less certain. Sending cold messages, publishing content, trying communities, and checking whether people truly have the problem are harder in practice than the advice sounds. The central challenge is not only making the product. It is getting the first person to trust the offer enough to pay attention or become a customer.
OBPrint is an online management product for print shops. It handles quotes, orders, production tracking, payments, and AI marketing tools. A short two-minute demo could have shown every feature one after another, but it did not. It starts with one clear moment: a shop owner standing in the business and not knowing where anything is. That single problem makes the later features feel like answers, not just more screens. The main lesson is simple: the more a product can do, the more important it is to begin with one clear customer problem.
Tamkar is a small app for people who have finished work but still need to chase payment. It sends invoice reminder emails so the business owner does not have to keep writing uncomfortable follow-ups. The product lets users add their own payment link, so it claims to charge 0% transaction fees. The free plan allows 5 to 10 invoices per day, and the pro plan has a two-week trial with no credit card required. Setup and sending an invoice are presented as something that can be done in under a minute. The build uses React, Vite, Express, Supabase, Resend, and Postgres. The harder problem was not scheduling automatic reminders, but writing emails that sound polite and firm without sounding passive-aggressive or desperate.
Early-stage solo founders often weaken cold outreach by leading with the product instead of the buyer’s problem. Many pitches now start with AI, but the person who might buy may not care about the technology itself. A car lot owner is more likely to care about a CRM that fits the business than about AI as a concept. Even that need may not feel urgent, or it would probably already be getting fixed. Cold outreach has to speak to the person who can actually buy, in words that make them feel understood. The opening should be specific to the customer’s situation, not centered on features or an AI agent. A pitch that starts with a “new AI workflow agent” may not move someone who does not understand or care about that kind of tool.
An open-source social media scheduler is available to run on Cloudflare Workers. It can be deployed and tried in about one minute, without connecting separate file storage such as S3 or R2. It supports scheduled publishing to Bluesky, Discord, LinkedIn, Mastodon, Telegram, TikTok, X, and YouTube. The tool runs on Cloudflare Workers and D1, and it handles files up to 25 MB by splitting them into chunks. The frontend was made with vibe coding, while the backend was not.
Baery is an app for small bakeries that need one place to manage custom orders, prices, and customers. It was built around the problem that many small bakeries still use notebooks, spreadsheets, or several separate apps to handle inventory, production, and daily work. The target user is a cake or cookie business that takes orders through direct messages, texts, and phone calls instead of a storefront. The app includes a custom order manager with an intake form, order list, and activity log. It also includes a recipe cost calculator with cost breakdowns and a margin slider, plus inventory and waste tracking and a lightweight CRM. Baery is available now on iOS, with a Google Play version planned. The official website is baery.app.
A dev tools launch missed search terms used by people who were close to buying. Searches like “X vs Y” and “best X alternatives” can matter because they happen near the final decision. Hiring a writer to create competitor comparison pages did not work well: only about 8 pages were finished in 3 weeks, the drafts felt generic, and review time became a major burden. The result was tryserpa.com, a tool that creates drafts for product-versus-competitor pages and roundup pages such as “best [category] tools.” The workflow lets the operator review a draft, change it if needed, or publish it. The claimed result is roughly 40 pages produced much faster, aimed at buyers with clear purchase intent.
Thinking Canvas is a research tool that puts Claude chats, webpages, PDFs, and notes on one open canvas. A person can keep source material and AI conversations side by side instead of switching between a browser, a notes app, and a chat app. Chat nodes can branch into separate lines of thought, so a side question does not need to interrupt the main research path. Chats can also connect to resources, such as rewriting a note, reading a webpage, or adding a PDF to project memory. Notes are stored as local-first Markdown files, so they can also be edited in tools like Obsidian. The tool is free to use, but Claude still creates token costs through an API key or a Claude subscription. Example uses include reviewing computer history, finding products that turn handwriting into digital text, and studying Reddit discussions for marketing copy. Similar tools include Obsidian Canvas, NotebookLM, Heptabase, and Slashspace, but Thinking Canvas focuses on combining canvas notes, Claude chat, local files, branching chats, and source connections in one place. The project has a website at thinkingcanvas.xyz and is open-sourced on GitHub.
ShuffleBall Arena is a casual browser game that mixes shuffleboard-style scoring with ideas from bumper pool, pinball, and Frogger. The next update is adding more visual and sound polish to make the game feel more like an arcade game. The main risk is that extra effects may make the board harder to read when several marbles are on it. The first design is cleaner and simpler, but players must read the score numbers to know what each target is worth. The second design gives each scoring ring its own color, makes rings glow when a marble lands in them, and adds rotating lights to some rings. That makes high-value targets easier to spot quickly, but it may also create too much visual clutter during play.
A small web project made for two friends who enjoy tier lists reached 40 to 90 daily users after a couple of weeks. The product gives people one tier list per day, so the activity stays simple and repeatable. It began as a fun tool for friends, but people outside that small group are also using it. A similar small voting tool found that showing the group average right after someone votes can give people a reason to come back the next morning.
Builders are invited to show real projects they are working on now, including a Micro SaaS, MVP, landing page, Chrome extension, bot, internal tool, AI wrapper, marketplace, dashboard, or unfinished idea. The work does not need to look polished. Early versions can show the real product thinking more clearly: the problem being solved, what is already easy to understand, what still feels unclear, and where the product might go next. The feedback focuses on what a regular user understands right away, what causes confusion, what seems useful, where the positioning is unclear, and what could be improved. Uplinkly started as a tool for branded links and QR codes, then expanded toward campaign and traffic tools such as landing pages, traffic flows, postback, routing, and analytics. Building it shows how hard it can be to explain a product in a simple, clear way.
Small software founders often build research or lead-finding tools before they know what task truly repeats. A better first step is to write down the exact words a buyer uses when the problem appears. Those words can then be searched on Reddit, Google, and LinkedIn, with 20 real examples saved for review. Before building an alert or dashboard, the founder should reply, help, and learn by hand. Only the part that repeats clearly should become a product feature. If the work cannot be done manually 20 times, the product idea may still be too vague. If it can, the MVP becomes clearer: what goes in, what comes out, who wants it, and what they do next.
Sprintools.dev is being built as a toolkit for founders who have not built a product yet or are still very early. It aims to put market research, lead finding, B2B outreach writing and automation, marketing content, and legal document generation in one place. The legal documents include NDAs, terms of service, and basic contracts. The goal is to replace four or five separate tools that often cost about $20 to $50 per month each. The product is not live yet; it currently has a landing page, a waitlist, and core features in development. The main pricing question is whether one monthly subscription makes sense, or whether the bundle lowers sign-ups because many buyers only want one or two of the five features. Another question is whether flat monthly pricing or usage-based pricing works better for very early founders who may not have revenue yet. The bigger positioning question is whether “everything for founders” is too broad, and whether starting with one strong wedge feature would sell better before expanding.
PostPeer.dev, a social media posting API, reached $1,566 in MRR about three months after launch. MRR grew from $34 in April to $402 in May, $1,270 in June, and $1,566 at the start of July. The product has more than 880 users and 79 active subscription customers. Total orders are 152, including one-time purchases, and total revenue is $3,574. It has 7 five-star reviews. The main growth drivers were replying to support requests within minutes, using SEO through how-to pages, comparison pages, and free tools, and building in public on Reddit and LinkedIn. The next experiment is selling to AI agents. MCP and agent skills now let tools like Claude use the API directly to write, schedule, and publish posts.
For a solo SaaS business, building the product can feel easier than getting people to notice it. If AI helps with marketing, one key choice is whether it should reply to people online by itself or only find people already talking about the same problem so the founder can respond personally. For content, AI could suggest proven post formats that are already getting attention and may give new posts a better chance of spreading. For analytics, simple advice may be more useful than raw counts like likes, comments, and points. The core question is whether AI should run marketing from start to finish or only handle dull repeat work while the founder still does the talking and thinking.
TrustMRRR, a SaaS product, reached about $6,000 in MRR after 15 months. The main lesson is to prove that people will pay before spending a long time building. A simple landing page with a Stripe payment button can test whether buyers exist before the product is finished. If the product can be built in two weeks, building first may be fine; if it will take longer, selling first is safer. Real payments give stronger proof than surveys because they show both demand and the ability to find customers. Growth came from sharing meaningful product updates, recording app changes, posting every extra $1,000 of revenue on X, and explaining lessons learned at each step. Building and marketing need separate time blocks because making the product is often easier than getting people to notice it. Content should be shared across X, Threads, LinkedIn, Bluesky, and Mastodon, while longer-term channels like SEO and GEO should be built after validation because social media algorithms can change. Short, natural talking-head videos can work even when they are not polished, and AI-written scripts are discouraged. Keeping stable income during the early stage can reduce stress and burnout, while helping other founders can create relationships and future opportunities. The first customers came from posting updates on X, engaging with people, and keeping a product link in the profile and pinned post.
A first SaaS usually does not need separate services for the front end, back end, and database on day one. A small product can often run on one VPS, or on a simple PaaS setup using services such as Railway, Render, Vercel, Netlify, and Supabase with managed Postgres. The low-cost VPS route often means using providers such as Hetzner, Contabo, or Webdock, then running Docker, Caddy, Coolify, or Dokploy to manage the app; examples ranged from about 5 to 30 euros, or 10 to 30 dollars, per month. The easier PaaS route reduces setup work, but the real cost should include the paid tier after the free plan, especially for the database. Cloudflare was often used as a free DNS and protection layer, and apps with user uploads need to check object storage and download costs. The two things worth paying attention to from the start are database backups that have actually been tested and a simple Git-based deploy path. Scaling can wait until a real bottleneck appears; before that, a bigger server is often enough.
TuringShot is a small macOS screen-effects app. Its early roadmap tried to catch up with larger tools by matching their feature lists and then adding one extra feature. That was a weak fight for a solo maker, because funded teams can ship more, and having the same feature rarely makes people switch. The useful shift came from noticing that competing tools create value at the same time: people record first, then add zooms, highlights, and callouts during editing. TuringShot changed its positioning so the emphasis happens live, while recording or presenting, instead of afterward. The product stayed in the same category, but the value moved to a different moment. That made it less of a feature-by-feature comparison and more of a clear reason to exist. Solo makers can look for a different axis, such as timing, audience, or workflow, instead of chasing feature parity forever.
Causo turns poorly targeted cold outreach into a chance to pitch its own product. The team calls the tactic “Uno Reverse sales.” If an office broker asks about a lease, the reply says the company is fully remote, but Causo can help find companies with real office space and estimate their space needs and rent. If someone mistakes Causo for customer support, the reply says it is not support, but it helps startups research prospects and write outreach with the right context. If an executive assistant agency assumes the company has raised money, the reply says it has not raised, but the product helps startups find and contact the right prospects. The logic is simple: the sender already does outbound, their targeting or context is weak, and they may understand the pain of finding good prospects. Instead of ignoring the message, the team tries to turn it into a useful sales conversation.
SubHunt is a Reddit-related SaaS product. Its operator can no longer use PayPal because of a business name change issue on the PayPal business account. Stripe is not available in the operator’s country, so a common backup payment option is not available either. SubHunt is now open to purchase offers. The service uses Netlify, Resend, and Supabase.
A solo technical founder has spent about two years running SaaS apps for e-commerce merchants. The apps are sold through one regional platform’s app store. The business has 81 subscriptions and about $800 in monthly recurring revenue. The founder needs about $1,400 per month to cover living costs. Net subscriber growth over the last four months was 2, 7, 8, and 11, so growth has been improving. Savings can cover about 12 more months, and the business may reach personal break-even in 9 to 12 months if the current pace continues. The biggest risk is that about 95% of revenue comes from one app on one platform. The practical decision is whether to keep pushing, take freelance work or a job for cash, or trust that growth will compound from this small base.
Growing software companies need to keep track of changing regulations and judge how those rules affect their products and operations. A key issue is ownership: compliance may sit with engineering, legal, security, or a dedicated compliance team. New rules or audits can force urgent changes to code or infrastructure. The time-consuming work is spotting rule changes, turning them into concrete product work, and preparing evidence for audits. The most useful automation would likely reduce that repeated tracking, mapping, and evidence-gathering work.