18:08
How I Turned Claude Into My Personal Assistant (Full Guide)
AI Edge
Overview
This video details the creation of a personalized AI-powered operating system designed to manage tasks, finances, and daily life. The creator, Miles Deutsche, explains how he moved beyond generic SaaS tools to build a custom system using Claude, Superbase, and Telegram. The system features voice note integration for task logging, automated prioritization, financial tracking, habit monitoring, and a journaling function that trains the AI on personal patterns. The goal is to create a highly personalized, efficient, and adaptable system that acts as a central hub for managing one's life and improving productivity.
How was this?
Save this permanently with flashcards, quizzes, and AI chat
Chapters
- Traditional task management systems (like Notion) failed due to manual data entry, leading to missed tasks and lost opportunities.
- A specific missed investment opportunity in Anthropic cost over $1.2 million due to disorganization.
- The vision is to move from generic SaaS to purpose-built AI tools that perfectly match individual needs.
- The creator built a custom AI operating system to manage tasks, finances, and calendar, automatically prioritizing the day.
Understanding the pain points that led to this custom system highlights the limitations of existing tools and the potential benefits of a personalized AI solution.
Missing an investment opportunity in Anthropic due to disorganization, resulting in over $1.2 million in opportunity cost.
- The system uses a front-end dashboard that reflects a back-end memory system, which can be transported to any LLM.
- Voice notes sent via Telegram are transcribed by Whisper, categorized by AI, and logged into a Superbase database.
- The dashboard automatically updates and prioritizes tasks based on the logged information.
- Financial data is pulled live into a Google Sheet and displayed on the dashboard for real-time net worth tracking.
This outlines the technical foundation of the system, showing how different AI and database technologies are integrated to create a seamless user experience.
Sending a voice note via Telegram about a new client proposal, which is then automatically transcribed, categorized, and added to the task management system.
- The design process began with mockups in Claude design, followed by exporting to Claude code for functional development.
- Superbase was chosen as the memory system for its sovereignty and scalability, acting as the 'brain' for data storage.
- An API key from Anthropic was obtained, and the system was built on Claude, though it could also be built on models like GPT-4.
- A Telegram bot was set up via BotFather, connected to a Vercel function via Claude code, to handle voice note transcription and logging.
This section details the specific tools and steps involved in building the system, providing a roadmap for others who wish to replicate or adapt it.
Using Claude code to convert a design mockup into a functional website architecture.
- Google Calendar is integrated for live updates, allowing users to see upcoming meetings directly on the dashboard.
- A habit tracking feature allows users to check off daily habits, with completion of subtasks marking the main task as done.
- A journaling function allows users to voice note their daily reflections, which are stored and can be analyzed by the AI for patterns.
- This daily journaling trains the AI on personal thinking patterns, enabling it to act as a coach or mentor.
These features demonstrate how the AI system goes beyond simple task management to support personal growth, self-reflection, and habit formation.
Voice noting daily reflections into the journaling feature, which the AI later analyzes to identify patterns like the link between sleep and productivity.
- The CRM module was the most challenging to build, requiring extensive debugging to ensure data integrity.
- Nutrition tracking allows logging meals with AI-estimated calorie counts, integrated into a health tab for daily intake monitoring.
- The finance section connects to a Google Sheet for real-time net worth and income tracking.
- The system is designed to be adaptable, with features like nutrition tracking being optional and customizable.
These advanced modules show the system's capability to manage complex personal data, offering insights into health, finances, and customer relationships.
Logging a meal of chicken and rice, which the AI estimates at 550 calories, and seeing the daily protein and calorie intake on the health dashboard.
- The dashboard features a 'Finance Pulse' for net worth tracking, 'Key Tasks' prioritized daily, and 'Daily Habits' with subtask completion.
- A 'Creative Session' module encourages morning creativity, followed by community engagement and finance checks.
- The 'Evening Wind-down Routine' includes journaling, which trains the AI and provides daily summaries.
- The system is designed for constant use, with a mobile-optimized web app for on-the-go access.
This provides a practical demonstration of how the creator uses the system daily, illustrating its functionality and integration into a productive workflow.
Starting the day by reviewing 'Key Tasks' on the dashboard, which are the 3-5 most important items selected from the CRM section.
- The core value lies in the back-end memory system, which is transportable to any AI interface (Claude, ChatGPT).
- This custom memory system provides exact context for AI interactions, unlike generic chat interfaces where memory is a black box.
- The system allows users to own and control their data and AI interactions.
- While the dashboard can be redesigned, the back-end memory is the true 'magic' and foundation for future iterations.
This emphasizes the long-term value and flexibility of the system, highlighting how it empowers users with control over their personal data and AI interactions.
Prompting the AI with 'what are the top three things I should do right now?' and receiving strategic recommendations based on the entire history of logged data.
Key takeaways
- Generic SaaS tools often fail to meet individual needs, driving the demand for personalized, purpose-built AI systems.
- Integrating voice input (like Telegram voice notes) significantly reduces friction in logging tasks and information.
- A robust back-end memory system is crucial for AI personalization and allows for data portability across different AI models.
- AI can be trained on personal data through journaling and daily reflections to provide tailored coaching and insights.
- Building a custom AI operating system requires iterative development, debugging, and continuous refinement based on user experience.
- The ability to automatically prioritize tasks and track key metrics like net worth and habits dramatically enhances productivity and organization.
- Owning and controlling your personal data within an AI system is paramount for privacy and effective AI interaction.
Key terms
AI Operating SystemSaaS (Software as a Service)Purpose-Built ToolsClaudeSuperbaseTelegram BotWhisper (AI)LLM (Large Language Model)API KeyCRM (Customer Relationship Management)Habit TrackingJournalingNet Worth TrackingTransportable Memory System
Test your understanding
- What were the main limitations of traditional task management systems that the creator aimed to overcome?
- How does the system leverage voice notes to automate task logging and prioritization?
- Why is a 'transportable memory system' considered the core value of this AI operating system?
- What role does the journaling feature play in training the AI and providing personal insights?
- How can building a custom AI system offer advantages over using off-the-shelf productivity tools?