Evolving my personal voice note app with MCP ✨. Now I can go for long walks on the beach… blabber away. Then Claude/Cursor can pick them up and transform it into docs, insights, content, etc.

Built my first MCP for a client’s ecom app, the unlock is insane. The objective was simple…. Give stakeholders complete transparency into the data and the ability to do their own analysis (without requiring additional dev work or humans). With this they can essentially generate.

Simple prompt hack when building in cursor / claude code. Add. "Does that make sense?" to the end of your instructions, feedback, context, etc. This forces your AI partner to repeat back what it has understood about your instructions and expectations. It's a quick way to.

Because the world needs another voice note app.

My biggest learning from building software with AI for the last 6 months (as someone who "can't code"):. You're not limited by technical knowledge or syntax anymore. What actually matters:. 👉 Ability to articulate requirements + desired outcomes. 👉 Ability to gather and.

Claude Code is really the best. Helped me build a Swift app built in 2 hours. I'Ve always been intimated by XCode. But because Claude code is CLI it makes it feel almost effortless. Transcription app with:.- faceID.- openai transcription.- queues.- background mode.-

I'm looking for senior-level fullstack engineer who is AI is leveraged and know's how to rip ⚡️. Bonus points if you have some design chops. I'm heading up product for startup who's building a suite tools/offering for their high ticket community. Lots of variety and what we are.

My AI Coding Workflow (that allows me to run multiple dev projects as a solo "vibe" studio). -------------.8:00am - ChatGPT Voice Mode.Kiddo drop-off. Voice brainstorm PRDs, bugs, implementation approach while driving. 8:30am - Codex Kickoff.Drop in PRDs/tasks, run automated.

Let's see if I can get all my work done for the week before Monday hits 🤞🤖👀

This approach has completely transformed what I'm able to achieve and build. IMO they key is to act more like a PM - AI excels at execution, but needs proper context and guardrails to stay focused and deliver exactly what you expect.

🧪 Testing & Deployment. After completing implementation, I test everything thoroughly in my local environment first. Once I'm confident locally, I deploy to staging for another round of testing before finally deploying to main for production. Currently I do a lot of manual.

📓 Documentation. As you complete tasks, Taskmaster maintains a running record of everything you've done, current status, and what's next. When the feature is complete, you have a comprehensive breakdown - perfect for team sharing, code reviews, and future reference if you need.

🚧 Implementation. After thorough review/iteration on the taskmaster plan, I'll create an official feature branch and begin working through the task list Taskmaster generated. The beauty of Taskmaster is its flexibility - easily adapt the plan, add context, and move tasks as.

✅ Planning Part 2 w/ Taskmaster MCP. Recently been using Taskmaster MCP and it's incredible. It breaks everything down into tasks and subtasks. You run the parse PRD command and it will break down your project into tasks and subtasks. The cursor agent will use this to.

🧭 Planning Part 1 w/ o3. I dump the context and PRD into ChatGPT o3 for implementation planning. Different AI tools have different strengths - Gemini excels at digging through codebases, while ChatGPT o3 is great for planning. The next step involves taskmaster MCP so I give o3.

📝 Draft a simple PRD. I don't over think this. I record myself talking through we need to build, what are the functional requirements, objectives, and pain points we're solving, what I imagine the user flow will be, etc. Basically massive brain dump. When working with AI, think.

📚 Gather context. Use Cursor + Gemini 2.5 Pro to dig through the codebase first. You need to understand the affected components, map relevant files, and analyze intersecting functions before writing a single line. Context building within your codebase is key - this enables you.

How I vibe code new features on large, existing projects. I've been vibe coding full-time for the last 6 months. IMO - to be an effective vibe builder, you are 60% PM, 25% technical architect, and 15% designer (this varies but rough breakdown). Here's what works for me.

Early draft of a highly custom piece of software for an attorney with a solo practice. Productivity app (Chrome extension) that runs entirely in Google Workspace. Completely optimized for their specific workflow:. 📬 Deep Gmail/Calendar/Drive/Docs integration.📂 Automated project