Step-by-step Claude workflows from How I AI expert interviews. Each guide includes prompts, code snippets, and practical tips.
View all How I AI episodes
Improve your AI's performance by moving from a single, massive context file to a modular library of small, focused markdown files. This system allows the AI to intelligently load only the most relevant information for each task, enabling you to use simple, 'lazy' prompts and get better results.
Build a custom research assistant that automatically finds relevant academic papers and generates high-quality summaries. This system delivers a daily research digest to your to-do list, helping you stay on top of your field with minimal effort.
Escape the friction of traditional GUI-based tools by building a bespoke task management system that runs in your terminal. This workflow uses Claude Code and markdown files to create and manage your to-do list with simple slash commands and natural language.
Go from an empty folder to a playable 3D multiplayer prototype in about 15 minutes. This advanced workflow leverages the Claude command-line interface and parallel development to rapidly build and debug both the frontend and backend of a flight simulator game.
Learn to build a complex 3D multiplayer game, even with no prior game development experience. This workflow guides you through an iterative process of using an AI assistant like Cursor and Claude as a junior developer to go from a simple concept to a functional prototype.
Use Repo Prompt to gain surgical control over your AI's context. This workflow shows you how to manually select specific code files, combine them with a detailed prompt and a persona, and generate a highly-contextualized input for an LLM like Claude or Gemini.
Create a powerful personal AI assistant by securely connecting your WhatsApp messages to a large language model like Claude. This workflow uses an open-source Model Context Protocol (MCP) to allow your AI to read your recent messages, summarize conversations, and even send messages for you.
Create a reusable system for generating high-quality, on-brand job descriptions using Claude Projects. This few-shot prompting technique saves time and ensures brand consistency for all hiring managers.
Take a detailed product blueprint and use the Replit AI agent to build, iterate, and deploy a complex, database-driven web application. This workflow demonstrates how to guide the AI through conversational prompts to create a functional app without deep coding knowledge.
Learn to transform a simple application idea into a professional-grade development blueprint. This workflow guides you through using Claude to generate detailed user stories, define system architecture, and create a phased rollout plan without writing any code.
Create a personalized AI brainstorming partner to help you prepare for high-stakes meetings and navigate complex stakeholder dynamics. This workflow uses Claude and classic literature to generate sophisticated communication strategies and actionable meeting plans.
Learn how to prototype and build a production-ready application to automate the creation of massive FDA submission documents. This workflow takes you from a simple prompt in Claude to a functional Streamlit app with features like PHI redaction and cost analysis.
Use AI to analyze test logs, identify and prioritize technical debt, and generate a task list for AI agents like Cursor or Devin to systematically fix issues. This creates a repeatable process for improving codebase health.
This workflow shows you how to design a custom categorization system, like a Dewey Decimal System for any collection, and use AI to automatically classify hundreds of items in an Airtable database, saving countless hours of manual work.
Learn how to use a large language model like Claude as a central 'brain' for your new business. This workflow covers generating financial projections, understanding local permits, and developing customer personas from a single conversational interface.
When a single AI model is stuck or not providing the output you want, use this simple but effective technique. Make two different AI models compete against each other to produce a more accurate and refined result.
Elevate your Midjourney results from generic to professional quality using the Subject-Setting-Style framework. This workflow teaches you how to structure your prompts with specific photographic language, such as film stocks and camera metadata, to create curated, high-end images.
Move beyond 'vibe prototyping' by first generating a structured JSON data model with an LLM like Claude. This workflow demonstrates how to create a rich, realistic dataset with real media, which you can then feed into a prototyping tool for a more accurate and flexible UI.
Offload the mental burden of context switching by creating a dedicated AI assistant in Claude or ChatGPT. This 'second brain' holds all your project context and can even scrape and analyze thousands of customer conversations from sites like Reddit.
Build a highly efficient, AI-powered personal operating system by centralizing all your business context—notes, data, and prompts—into a GitHub monorepo. This allows AI assistants like Claude or Gemini to consistently and accurately automate complex tasks like writing proposals and creating content.
Design and refine a conversational AI by starting with ideal user dialogues in Claude. This workflow takes you from writing sample conversations to generating a fully interactive, LLM-powered prototype with a system prompt.
Move your custom Claude Skills from your local development environment to the user-friendly Claude.ai web interface. This guide shows you how to package your skill folder, upload it, and begin using it for daily tasks directly in your browser.
This workflow demonstrates how to use a previously created 'meta-skill' within the Claude Code command-line interface to generate a new, specific skill. You'll learn how to prompt Claude Code to build a skill and then immediately test it.
Learn the author's preferred method for building a powerful 'meta-skill' inside the Cursor editor. This workflow uses Anthropic's official documentation to create a reusable agent that can quickly generate new, well-structured Claude Skills on demand.
Use Spec Kit with an AI assistant like GitHub Copilot to improve the quality of your software specifications. The AI acts as an intelligent partner, analyzing your requirements and asking clarifying questions to help you identify edge cases and ambiguities before development begins.
Develop a custom iOS app like 'Flip Flop' to streamline research in physical archives. Use AI to automatically describe and transcribe photos of documents and embed the data directly into the image files.
Transform dense research documents and buyer personas into interactive, conversational AI agents using NotebookLM and custom GPTs. This allows your team to directly query customer personas for real-time insights.
Create a custom AI application that connects to your calendar and email via APIs to help you strategically review your day, delegate tasks, and triage your inbox with brutally honest feedback.
We've made ChatPRD affordable so everyone from engineers to founders to Chief Product Officers can benefit from an AI PM.