All activity
Monk Modeleft a comment
I use both and they solve different problems. Cursor (IDE): Best for when I'm working inside a file and need inline completions, quick refactors, or want to chat about a specific function. The tab completion is addictive once you get used to it. Great for "I know what I want to change, help me do it faster." Claude Code (terminal): Best for bigger tasks. "Create a new API endpoint with tests...
AI in your IDE (e.g. Cursor) vs AI in your terminal (Claude Code) — what’s the better flow?
Aaron O'LearyJoin the discussion
Monk Modeleft a comment
Both. And I think that's the right answer for most people building in 2026. I recently shipped a native Mac app (Swift + SwiftUI). The core architecture, data layer, and performance-critical code I wrote "the old way" because AI still struggles with complex state management in native apps. But for boilerplate, UI layouts, and utility functions? Claude and Cursor handled probably 40% of that....
Building SaaS in 2026: Are you vibecoding your own product or engineering it the "old way"?
Aleksej VukomanovicJoin the discussion
Monk Modeleft a comment
Usage-based is becoming the norm for sure, but I think there's a backlash brewing. Developers are getting tired of unpredictable bills. The problem with usage-based pricing for AI tools specifically is that usage is really hard to predict. You write a prompt, the model decides how many tokens to use for the response. You don't control it. Multiply that by hundreds of API calls per day across...
Is usage-based pricing becoming the norm for AI tools?
Jake FriedbergJoin the discussion
Monk Modeleft a comment
Late to this thread but sharing anyway since accountability helps: Q1 goals were: 1) Ship the side project I'd been sitting on for 6 months, 2) Get to $500 MRR equivalent on my indie apps, 3) Actually track where my money goes instead of guessing. Progress: Shipped the project (a Mac utility for tracking AI API costs called TokenBar - tokenbar.site). Didn't hit $500 yet but learned a ton about...
2026 and your goals. Let's try to set them for Q1.
Monk Modeleft a comment
Unpopular take: don't look for a co-founder. At least not right away. I spent months trying to find the "perfect co-founder" for my last project. Went to meetups, did the YC co-founder matching thing, had dozens of coffee chats. None of it worked because you can't evaluate someone you've never actually shipped something with. What worked: I started building solo. Launched an MVP. Got early...
How do you find co-founders (and figure out if they’re a good fit)?
Monk Modeleft a comment
For me it comes down to three things: a clear one-liner that tells me exactly what it does, a demo or screenshot that shows it in action, and pricing that makes sense. The launches that lose me are the ones where I have to read three paragraphs just to understand what the product does. If I can't get it in 5 seconds, I bounce. Also - and this might be controversial - I'm way more likely to...
What makes you click into a Product Hunt launch?
Monk Modeleft a comment
Building TokenBar (tokenbar.site) - a native macOS menu bar app that tracks your AI spending across 20+ providers in real time. Stack: Swift + SwiftUI for the entire app. No Electron, no web views, fully native. Uses each provider's API to pull usage data and processes everything locally on the user's machine. Zero cloud infrastructure needed which keeps costs at basically zero. The idea came...
What are you building, and what does your stack look like?
Ryan HendricksonJoin the discussion
Monk Modeleft a comment
For me the answer came down to one simple rule: if the feature creates immediate value, it should be in the paid tier. If it builds trust and lets people understand the product, keep it free. I built a Mac menu bar app called TokenBar that tracks AI spending across 20+ providers. The free/paid split was tough but I landed on: Basic tier ($5 one-time) gives you real-time spending tracking for up...
How do you decide what features should be free and what should be paid?
Monk Modeleft a comment
I use both and they serve different purposes. Cursor (IDE): Best for when I am editing existing code. Inline completions, cmd+K for quick edits, tab-tab-tab through suggestions. It feels like pair programming. Great for iterating on UI code, fixing bugs, and refactoring. Claude Code (terminal): Best for greenfield work and multi-file changes. When I need to scaffold a new feature, create...
AI in your IDE (e.g. Cursor) vs AI in your terminal (Claude Code) — what’s the better flow?
Aaron O'LearyJoin the discussion
Monk Modeleft a comment
Security was the #1 concern when I built TokenBar (a menu bar app that handles API keys for 20+ AI providers). Here is what I did: 1. Everything runs locally. No cloud, no backend, no accounts. Your API keys never leave your Mac. This eliminates the entire class of server-side vulnerabilities and data breaches. 2. macOS Keychain for secrets. API keys are stored in the system keychain, not in...
How are you dealing with vibe coding security risks in AI-generated code?
Monk Modeleft a comment
Both. I vibecoded a native macOS app (TokenBar, tracks AI spending across providers) and the answer is really "it depends on the layer." Architecture and data flow: old way. I designed how the app talks to 20+ different AI provider APIs, how data gets cached locally, how alerts trigger. Claude is not great at system-level decisions that require understanding the full picture. Implementation:...
Building SaaS in 2026: Are you vibecoding your own product or engineering it the "old way"?
Aleksej VukomanovicJoin the discussion
Monk Modeleft a comment
For native macOS apps, my stack is: Swift + SwiftUI + Claude (API and Claude Code in terminal). No web frameworks, no Electron. I built TokenBar (a menu bar app that tracks AI token usage across 20+ providers) with this stack. Claude handles the boilerplate, SwiftUI layout, and especially parsing different API response formats. I still write the core logic by hand, but Claude probably handles...
What’s Your Vibe Coding Stack in 2025?
Monk Modeleft a comment
As someone who launched a small utility app (TokenBar, a macOS menu bar app for tracking AI spending), here is what I have learned about what makes ME click into launches: 1. A clear, specific problem in the tagline. Not "revolutionize your workflow" but "track your AI spending across 20 providers from your menu bar." I want to know exactly what it does in under 10 words. 2. A screenshot or GIF...
What makes you click into a Product Hunt launch?
Monk Modestarted a discussion
Solo dev building a Mac menu bar app to track AI spending. Happy to be here!
Hey everyone! I am a solo developer who has been deep in the AI tools world for the past couple years. I use Claude, ChatGPT, Cursor, and a handful of APIs daily for my work. One month I looked at my credit card statement and realized I had spent $340 on AI tools without noticing. Every provider has its own dashboard, its own billing page, its own way of showing usage. I never checked any of...
Monk Modeleft a comment
Building TokenBar, a macOS menu bar app that tracks AI token usage and spending across 20+ providers (OpenAI, Anthropic, Cursor, Gemini, Copilot, OpenRouter, etc.). Stack: Swift + SwiftUI, native macOS. No Electron, no web views. Everything runs locally on your Mac. I chose native because menu bar apps need to be lightweight and fast. Nobody wants a 200MB Electron app just to show a number in...
What are you building, and what does your stack look like?
Ryan HendricksonJoin the discussion
Monk Modeleft a comment
Trust depends entirely on what the tool is doing with your data. I built a Mac menu bar app (TokenBar) that tracks AI spending across providers, and the single most important design decision was making it fully local. No cloud, no accounts, no data leaving your machine. Your API keys and usage data stay on your Mac. I think a lot of AI tools get trust wrong by defaulting to cloud-first when...
How much do you trust AI agents?
Monk Modeleft a comment
For my app (TokenBar, a macOS menu bar utility that tracks AI token usage), I went with a simple one-time purchase. $5 for Basic, $10 for Pro. No free tier, no subscription. My reasoning: the app solves a clear, specific problem (you are paying for 5+ AI tools and have no idea what you are actually spending). If someone has that problem, $5 is an easy yes. If they do not have that problem, a...
How do you decide what features should be free and what should be paid?
Monk Modeleft a comment
My first vibecoded project was TokenBar. I was paying for Claude Pro, ChatGPT Plus, Cursor, and hitting a bunch of APIs on top of that. Checked my credit card one month and realized I had spent $340 on AI tools without noticing, because each provider has its own dashboard and I never checked any of them regularly. So I built a macOS menu bar app that shows token usage, credits, and billing...
What was the very first project you vibecoded with AI?
Monk Modestarted a discussion
I was spending $340/month on AI tools without realizing it. So I built TokenBar.
I pay for Claude Pro, ChatGPT Plus, Cursor, and hit various APIs on top of that. One month I checked and realized I'd burned through $340 on AI tools without noticing. Each provider has its own dashboard and I never checked any of them regularly. So I built TokenBar. It sits in your macOS menu bar and shows your token usage, credits, and resets across 20+ AI providers in one place: OpenAI,...
Monk Modestarted a discussion
Big update: 20+ AI providers, usage alerts, and new Pro tier
Hey everyone! Wanted to share what's new with TokenBar. We just shipped a major update that adds support for 20+ AI providers including OpenAI, Claude/Anthropic, Gemini, Cursor, GitHub Copilot, OpenRouter, and Vertex AI. You can now see all your AI spending in one glance from your Mac menu bar. What's new: Usage alerts that notify you when spending jumps unexpectedly (catch runaway scripts...