The Simple 4-Step Process To Build Your Own AI Trading Assistant With Claude (for Beginners)

Key Concepts

• Claude Code: An agentic AI tool that can write, edit, and execute code, as well as interact with local files and computer systems.
• Agentic AI: AI systems capable of performing tasks, making decisions, and executing workflows autonomously or semi-autonomously.
• Plan Mode: A specific mode in Claude Code where the AI brainstorms and creates an implementation plan without modifying any files.
• Build Mode: A mode where the AI is granted permission to edit files and build the project based on the provided plan.
• Routine/Daily Ritual: An automated workflow scheduled to trigger at specific times (e.g., 4:15 PM) to prompt the user for data entry or performance reviews.
• Discretionary Trading: A trading style where decisions are based on the trader's judgment, experience, and analysis rather than purely automated algorithms.

1. The Four-Step Workflow for Building AI Tools

The presenter outlines a repeatable framework for building any custom trading tool, emphasizing that the dashboard itself is secondary to the process of building it.

1. Plan Mode: Initiate a session in Claude Code using "Plan Mode." The goal is to brainstorm and define the project scope. The AI does not edit files here; it creates a .md (Markdown) file containing a concrete implementation plan.
2. Build Mode: After saving the plan, start a new session in "Build Mode." Provide the AI with the .md plan. The AI then builds the "skeleton" of the application (in this case, an HTML-based dashboard).
3. Personalization: Once the skeleton exists, use iterative prompts to add specific features, tabs, or data points. This is where the user tailors the tool to their unique trading style (e.g., adding a "Tendencies" tab or a "Performance Coach" layer).
4. Routine Implementation: Use the "Routines" feature to schedule the AI to interact with the user at specific times. This ensures consistency in data entry and daily reviews.

2. Technical Implementation Details

• Dashboard Architecture: The trading assistant is a local HTML single-page application. This allows the user to open it in any web browser, ensuring data privacy and accessibility.
• Data Entry: Rather than complex API integrations, the user opts for a manual entry workflow triggered by a daily routine. The AI asks specific questions (e.g., "What trades did you take?", "Best ops of the day?"), which the user answers, and the AI updates the HTML file accordingly.
• Efficiency: To save on token usage, the presenter recommends using "Ask User Question" mode. This forces the AI to present multiple-choice options rather than engaging in long-form conversational text, significantly reducing the computational overhead.

3. Real-World Application: The Trading Assistant

The presenter built a dashboard that functions as a personalized "TraderView." Key features include:

• Performance Tracking: Tracks P&L, win-loss ratios, and trade grades (A+, A, B, etc.).
• Tendency Analysis: The AI acts as a "Performance Coach," scanning past trade write-ups to identify recurring mistakes (e.g., "no man's land sizing") or positive patterns.
• Playbook Integration: The dashboard stores the user's specific "checks in favor" and rules for different setups, serving as a centralized reference point.

4. Key Arguments and Perspectives

• Focus on Inputs: The presenter and his co-host emphasize that the AI is only as good as the data provided. High-quality, detailed trade write-ups are essential for the AI to provide meaningful coaching.
• Simplicity vs. Complexity: A major concern discussed is the risk of "getting in the weeds." The co-host warns against building overly complex systems that provide no real ROI. The goal is to use the AI to simplify decision-making in the "simple moments" of trading, not to create a flashy but useless dashboard.
• The "1% Rule": The video references a statistic that only 1% of retail traders succeed, framing the AI assistant as a tool to help traders avoid common pitfalls and improve their odds of success.

5. Notable Quotes

• "I'm not going to talk in theory. I'm not going to tell you AI is powerful. You know that. I'm going to show you the exact four-step workflow I use." — The Presenter
• "The idea is you're going to have the steps... that you can build anything." — The Presenter
• "It's all about the inputs... just because it can process doesn't mean it's going to be smart." — The Co-host

6. Synthesis/Conclusion

The video demystifies the process of building custom AI tools for traders. By following a structured, four-step workflow—Plan, Build, Personalize, and Automate—traders with zero coding experience can create highly personalized dashboards that track performance, identify behavioral tendencies, and enforce discipline. The ultimate takeaway is that the value of such a tool lies not in the technology itself, but in how effectively it is tailored to support the trader's specific process and improve their decision-making in real-time.