Brian Garland

i bought a house last year in Kansas City and immediately felt behind. the furnace filter was overdue, the gutters had six inches of oak leaves packed in them, and i had no idea when to winterize the hose bibs. every app i tried gave me a longer to do list. none of them told me what skipping something actually costs. so i'm building Kempt. it's a home maintenance app for first-year homeowners. you snap photos of your house and your appliances, it builds a profile with real part numbers and real dates. it sends a weekly Sunday Brief instead of a pile of push notifications. and it keeps a running tally of the money you've saved by doing the small stuff on time. iOS beta is on TestFlight now. solo founder, bootstrapped, building in public. if you just bought a house and feel underwater on upkeep, that's who this is for. i'm Brian. nice to meet you.

wrote an App Store submission runbook today. every screenshot, every metadata field, every string. building solo means half the job is the product. the other half is making sure future you doesn't have to re-learn how to ship it.

@specsycoder @yashmp2004 kempt, an app to manage home maintenance. bottleneck has been deciding what features to build and what to not build. lots of apps out there doing similar things, all differently. want to differentiate myself building what people actually want

@UseKempt @yashmp2004 What are you building, and what's the real bottleneck?

Software Developer Roadmap to Get Hired in 2026 - Learn One Language Well → Java / Python / JavaScript / C++ -Master DSA → Arrays, Trees, Graphs, DP - Learn Git & GitHub → Version control is non-negotiable - Build Projects → Full-stack > Tutorial certificates - Learn Databases → SQL + basic NoSQL -Learn Backend → Spring Boot / Node.js / Django - Learn Cloud Basics → AWS, Docker, CI/CD - Use AI Tools → ChatGPT, Claude, Cursor - Learn System Design → APIs, caching, scaling - Build in Public → GitHub, LinkedIn, Portfolio -Apply + Network → Referrals > Cold applications -- Repeat for 6–12 months -- The harsh truth: 2026 won't reward the person who knows the most. It will reward the person who can prove they can build

this is what Kempt is built on. i bought a house and spent the first three months losing ground on maintenance while trying to figure out what i didn't know. solving one problem people will pay for is exactly right. the harder problem is knowing which problems are actually universal and not just yours.

You're more ready than you think if you can: • Solve one problem people pay for • Follow up without being annoying • Use AI to do the work of three • Handle hearing "no" a few times That's the starter pack. The rest is just reps.

the part that gets skipped in that conversation is the deferred maintenance that quietly erodes the asset while you're waiting for appreciation. gutters, hvac filters, the grading around the foundation. the stuff that doesn't show up in a Zestimate. appreciation is the goal but maintenance is the work that makes it possible.

Appreciation is not a strategy. Michael Zuber has watched people with $10M balance sheets go completely bankrupt betting on it. He's survived the Dot Com bubble, the 2008 crash, and the post-pandemic meltdown, and his rule hasn't changed: Hold for 10 years minimum. Buy for cash flow. Never count on appreciation. He's so confident in this market right now, he's pulling $1M out of his own properties to buy more. Do you factor appreciation into your deals, or do you treat it as a bonus if it happens? 👇 Listen to the full episode at youtu.be/ANG7XCcE_vc.

@ThisOldHouse once you go soft-close you can't go back. same with self-closing toilet seats. three months into home ownership and i've replaced every hinge in the house that didn't have one. the small stuff is what makes it feel finished.

🪵 Once you go soft-close, there's no going back. Nathan Gilbert walks through how to upgrade your cabinets so every door and drawer closes quietly and smoothly every single time.

@SkerdiDev writing clear requirements. i build an app with claude code and the gap is never the code. it's me not describing what i want precisely enough. a vague prompt gets a vague feature. a specific one with named screens and expected behavior gets something i can ship.

Developers… What’s the most valuable non-coding skill? 👇 • communication • product thinking • writing • sales • problem solving Which one made the biggest difference for you?

this is where i'm at right now. i'm a non-technical founder building an ios app with claude code. i can't write the code myself. but i know exactly what the app should do, who it's for, and which features matter. three months in, the product decisions are the bottleneck every single week. the code never is.

@yashmp2004 AI is making code cheaper. That means product thinking, taste, and understanding users are becoming more valuable. The industry isn't rewarding who writes the most code. It's rewarding who builds the best products.

this is where i live every day. the code part genuinely is solved for me. i describe behavior, Claude Code writes it, i test on my phone. what's still hard: knowing when the architecture is wrong before it bites you three weeks later. deployment configs. environment variables that work locally and break in production. the engineering judgment layer. AI doesn't have opinions about those things yet, so you still need to build that sense yourself, just faster than before.

the number of non-technical founders trying to build their own app has exploded vibe coding opened the door and people walked straight through it which is great, until they hit the wall almost every founder who has approached me recently has the same story they started building, got excited, then got stuck and it wasnt the code that stopped them ai writes the code, that problem is solved what stopped them was the software engineering layer underneath it all the architecture decisions, the tooling, the environments, the integrations, the deployment, the cloud infrastructure, the submission process and so on thats not something an ai prompt fixes coding is one piece of software development founders are starting to realise that and now they're coming to me not to build for them, but to learn how to navigate it themselves because they want to vibe code like an engineer, and thats the most encouraging thing i've seen in years

i'm a non-technical founder and Claude Code is where the real leverage is for me. i describe what each screen should do, Claude Code writes it, i test the behavior on my phone. shipping a full iOS app that way right now. cowork is great for document processing, but if you're building a product, Code is the one that changes what's possible for people who aren't engineers.

Most people are paying for Claude... ...and using about 10% of it. They open Chat. Ask a question. Get an answer. Repeat. That's not where the real leverage is. The biggest productivity jump comes when you stop treating Claude like a chatbot and start treating it like a coworker. That's exactly what this infographic breaks down 👇 Three parts of Claude that most users completely overlook: 🔹 Claude Chat Perfect for quick questions, research, brainstorming, and writing. 🔹 Claude Cowork The feature that changes everything. Instead of manually uploading files one-by-one, you point Claude at an entire folder and give it a job. 40 invoices? A month of reports? A folder full of client documents? Describe the outcome and let Claude work through it while you do something else. 🔹 Claude Code For developers who want Claude directly inside their terminal. But honestly? If you're not a developer, Cowork is probably the feature you should be learning next. One of the most valuable habits I've learned: Before running any major task, create three context files: ✓ Who you are ✓ How you work ✓ What you hate Claude reads them first. The result? Shorter prompts. More consistent outputs. Less repetition. Better results every session. The future isn't better prompting. The future is better context. That's where the biggest gains are happening. If you're still only using Chat, you're leaving a huge amount of value on the table. Join our network of builders and expand your AI knowledge: 🌐 stremit.io What feature are you using most today: Chat, Cowork, or Claude Code? #AI #Claude #ArtificialIntelligence #Automation #Productivity #AITools #FutureOfWork #BuildInPublic

@ConcertoRun @hellojaibu @X three months in as a solo, non-technical founder. i describe the full behavior for one feature, Claude Code writes it, i test on my phone before anything merges. batching parallel sessions is a different approach. curious how you handle review when multiple things land at once.

@hellojaibu @X hey! building Concerto, my fix for the solo-founder problem: instead of babysitting one Claude Code session, one chat lets Claude run a whole batch in parallel and report back. shipping like a team, alone. let's connect 🤝

Hey @X algorithom Trying to #connect with more people in tech/build in public 👋 Especially from: Web development Founders & Vibe coders AI/ML SaaS Startups Tech in general What are you guys building right now? Let's connect and grow together

@HomeZada a $20 furnace filter replaced on time saves $180 in compressor wear. that's the number most apps don't show you. they give you the list but skip the cost of ignoring it. the reminder alone doesn't change behavior. the dollar amount does.

Skipped maintenance doesn't just cost money. Loose deck railing → someone falls. Dirty air filter → air quality drops. Ignored smoke detector batteries → that's your life. A home calendar reminds you before problems compound. Ep 51 of Be a Smarter Homeowner 🎧 zen.homezada.com/2026/04/22/ep-… #HomeSafety #HomeIntelligence #HouseFacts

seasonal tasks and appliance tasks were overlapping in Kempt. you'd get a reminder to check your furnace filter AND a seasonal "prep your heating system" task in the same week. your house has a schedule. it shouldn't repeat itself.

@robinebers @clerk @authkit this issue may be the thing that gets me to start using Codex

dear claude code team i know i've been hard on you at times but can you PLEASE 🙏 just take care of the single biggest blocker that stops people from building apps with external login services like @clerk and @authkit? it is absolutely insane that codex can set up meta and google analytics, browse my youtube competitors, and e2e test my apps yet claude code, which has become such a great app overall, can't even run the simplest apps because of the localhost limit this is a true BLOCKER, not a hopeful feature request whoever is in charge over there, please fix this @trq212 @amorriscode @bcherny @_catwu this has been flagged thousands of times it's time

@mattwensing yep every once in a while I start to get worried about the costs of building a mobile app, but then I remember that I quite literally wouldn't have been able to do this before without spending tens of thousands of dollars. then those worries go away

a little shocked at how expensive it has been to play around with building a lil mobile app in 2026 claude code max $100 apple dev license $99 tokens $13.06 hosting $0.09 not a typo typing commands while watching the NBA finals: free crushing really

i know this feeling. non-technical founder, building a home maintenance app solo with Claude Code. no programming background. i describe what each screen should do, it writes the code, i test the behavior on my phone. three months in i have a working app on TestFlight. knowing what to build turned out to be the hard part. the code part has a solve now.

Technical non-technical founder 👩🏻‍💻😅 Even though I single-handedly designed the complete technical documentation - covering full logic, architecture, data normalization, CMS, and even the core algorithms, down to the copywriting and UX (which blew my lead developer’s mind to the point where he joined the project with zero questions, declaring full commitment and readiness to move to the US to work for me) - my biggest contribution to the project is still my elite sports background and medical expertise as a Medical University graduate. Up until now, despite everything I’ve built, I haven’t written a single line of code in my life. Even during our talks, my dev just accepted everything I proposed with zero objections. I’ve always had a deeply analytical mind, which honestly used to drive many of my tennis coaches crazy, testing their patience 🎾 😅 Maybe that’s why I understand the system well enough to design it down to the smallest detail, while at the same time... I just never learned how to program. However, we still don't have funding, and that’s blocking me big time - and I hate being blocked; it’s definitely not my natural state 😅 It feels like they expect me to build a fully finished system, run clinical trials, launch it to the market, and show MRR- just to give me money for... clinical trials and paying the team 😂 Am I the only one seeing a slight logical glitch here? 🤔 So, I decided to code the first MVP myself. Granted, we won't launch it to the public (since I need to outsource clinical trials first, and this is a human health management system, which is exactly what I need the funding for), but we are setting up a specific market experiment to show potential investors even better traction. We might even drop a pre-order option 😉🔥 I want to give maximum value to the people joining us, leveraging the fact that I already have the entire logic and full technical specification ready. So... I decided to code it myself 😎 If I don’t know how to do something, I just learn it 😉 It’s kind of funny that most technical people know programming languages inside out but have no clue what to build. I know exactly what to build - I have the full technical spec, a lead developer on the team, a scientist with over 110 publications to his name, and I’m constantly talking to the market, already holding 5 LOIs from world-class athletes across different countries 👌🏻 Now, all that’s left... is to guide the AI so it knows we are building the future 😎

the workflow is real but 2 hours undersells it. i describe what each screen should do, Claude Code writes it, i test the behavior on my phone. first version of a feature might take 2 hours. getting it right takes a week of back and forth. still faster than hiring a dev shop. but calling it easy would be dishonest.

What used to cost $50,000 and take 6 months now takes 2 hours and costs $25/month. This is how a solo founder out-builds a funded startup in 2026. For years, two blockers stopped people from building. "You need to learn to code." "You need $50K to hire developers."

@kidtsang Claude Code, Visual Studio. i'm a non-technical founder building a home maintenance app. my workflow is: i describe what each screen should do in plain english, Claude Code writes it, i test the behavior on my phone.

🚀 AI CLI: The Terminal-Powered Engine of Vibe Coding in 2026 AI CLI tools have emerged as a powerhouse in the vibe coding movement of 2026. These command-line interfaces let developers (and aspiring builders) describe projects in natural language—"vibe" prompts—and watch AI agents handle file reading, multi-file editing, debugging, testing, and deployment directly in the terminal. No fancy IDE required. Tools like Claude Code, Gemini CLI, Aider, and open-source agents such as Goose and Orion V2 are making terminal-native vibe coding faster and more powerful than ever. Latest X Buzz (June 2026): The vibe coding community is thriving with practical CLI applications. The STON.fi Vibe Coding Hackathon Wave 2 on TON blockchain is in full swing, with participants leveraging AI CLI agents for rapid TON/DeFi app development. Builders praise how CLI tools excel at handling complex, multi-file projects during sprints. Anthropic’s Claude Code dominates discussions as the go-to terminal agent. Users highlight its 200K+ token context for large codebases, on-demand file reading, and superior reasoning for refactoring. Free workflows combine it with VS Code, OpenRouter, or local models like Ollama—no API key always needed. One post shared a full setup for budget-friendly vibe coding. Trending Factors: Claude Code Leadership: Excels in complex tasks, codebase-wide changes, and agentic workflows. Often paired with Cursor for hybrid setups. 80% of Anthropic’s own code is now Claude-generated. Multi-Agent & Open-Source Surge: Goose (Block’s project, 45k+ GitHub stars) runs locally with any LLM for private prototyping. Orion V2 offers autonomous terminal agents with planning, subagents, and multi-provider support. Hermes ecosystem adds desktop integration, shared memory, and token optimization. Productivity & Accessibility: CLI tools shine for experienced devs comfortable in terminals—deep context, no visual bloat. Non-coders and side-hustlers use them for quick MVPs. Adoption stats show massive gains: 10x faster prototyping, with seniors reporting up to 81% boosts. Gemini CLI and others add multimodal capabilities. Stack Combos: Popular setups include Claude Code CLI + VS Code/Cursor, Aider for Git integration, and Lovable/Bolt for web-based augmentation. Budget options mix TRAE, Copilot Pro, and local LLMs. Challenges & Nuances:"Vibe coding hangover" persists—AI generates "almost right" code, leading to debugging debt and potential security issues. CLI’s text-only nature lacks visual previews, requiring terminal comfort and human oversight for production. Maintainability, Git workflows, and over-reliance are hot debate topics. Safety and guardrails remain critical as agents gain more autonomy. Implications & Edge Cases:AI CLI democratizes development: Non-technical founders build apps via prompts, while pros orchestrate sophisticated systems. In crypto and startups, "pivot to vibecoding" with CLI tools enables lightning-fast iteration and monetization. Edge cases include handling massive repos (CLI strengths), local privacy needs (Goose shines), and hybrid human-AI flows. Future trends point to self-healing agents, better evals, and seamless IDE-CLI integration. AI CLI isn’t replacing developers—it empowers them as high-level conductors. Whether refactoring legacy code, spinning up prototypes, or shipping in hackathons, the terminal has never been more alive. Clear prompts + powerful models + judgment = magic. What’s in your AI CLI vibe coding stack? Claude Code? Goose? Building something cool in the terminal? Share below! #ClaudeCode #GeminiCLI #Aider #GooseAI #AgenticAI #TerminalMagic #FutureOfCoding #TONHackathon #BuildInPublic #AICLI #DevLife

@naitikchame building Kempt, a home maintenance app for first-year homeowners. bought my first house and realized nobody tells you what your house actually needs or when. so i'm building the thing i wish existed. on TestFlight now. usekempt.com

If you are a solo founder, what are you building right now. Drop your product below 👇

@itbhakt building Kempt. it's a home maintenance app for first-year homeowners. tells you what to do, when to do it, and what it costs if you skip it. on TestFlight now. usekempt.com

The "What are you working on?" thread is back. 🧵👇 Founders, drop your SaaS below. I want to see what you're building and connect with you! 🚀 Tell me: 1️⃣ What it does 2️⃣ Who it's for 3️⃣ Link to your site RT to help a founder get some eyes on their product! 🔄