Phase 7a — Getting Opinionated: Rules-Based Auto-Categorization (and a Seam for the AI Later)
My expense app finally has a point of view on what I'm spending money on. No AI yet — just honest keyword rules, a nullable column, and one interface that means I can bolt an LLM on later without ripping anything out. Here's the build, three "empty value" bugs that bit me, and the habits that kept it clean. Index Where we left off The plan: rules first, AI behind the same door Step 1 — A nullable…
In the latest phase of the expense app development, a rules-based auto-categorization system is introduced, with provisions for integrating an AI categorization system later. The categorization process relies on a set of keyword rules and a nullable column in the database schema. Here is the step-by-step account of the development process:
1. A nullable category column is added to the Expense model. This column allows for expenses to be uncategorized initially, representing an authentic "no category" state. The nullable attribute is crucial for accommodating existing expenses created before the column was added, as migrating a non-nullable column to a nullable one would have caused migration failures.
2. The categorize() function is implemented to categorize expenses based on predefined keyword rules. This function serves as a reliable, offline fallback mechanism that can be easily swapped out with an AI categorization tool in the future. The categorize() function takes an expense description as input and outputs the corresponding category or "I don't know" if the rules cannot determine a specific category.
3. An interface is created to allow for easy integration of the AI categorization system later on without modifying the existing codebase. This interface ensures that the categorization process remains consistent and unrestricted, enabling the seamless swapping of the rules-based categorizer for an AI-powered categorizer.
4. The UI loop is designed to give users the final say in categorizing expenses, regardless of how accurate the auto-categorization becomes. This UI loop provides users with the flexibility to review and adjust the app's categorization decisions as needed.
5. The final step involves wiring the categorize() function into the create functionality, allowing users to input new expenses with associated categories. The categorize() function takes precedence over the AI categorization system during this process.
Throughout the development process, the author emphasizes the importance of building the boundary before plugging in external dependencies like AI categorization systems. By creating a rules-based categorization system first, the app will remain reliable and consistent, even as an AI-powered categorization system is integrated later.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — it may contain errors, so check the original before relying on it.
