Adam uses your business-specific instructions to categorize bank transactions intelligently. This guide walks you through accessing the training tools, creating effective rules, activating the banking agent, and understanding how Adam applies your custom logic.
Getting Started
To start training Adam:
Navigate to the Side Menu in the Adam dashboard.
Click on Train Adam.
On the Train Adam page, click Start Training.
You will see a list of AI agents, each specialized for different functions:
Revenue Manager
Expense Manager
Accounting Manager
Banking Manager (handles bank transaction categorization)
Reporting Manager
Financial Analyst
Each agent processes specific types of data to tailor Adam’s responses to your business. For bank transactions, select Banking Manager to customize categorization.
Setting Up Rules and Category Names
Define how Adam should interpret and categorize your transactions by adding custom rules:
Click the edit icon next to Banking Manager.
Add or update instructions in clear, simple terms. Use actionable language to specify criteria for categorization.
Example Instruction:
"Any transaction with description 'Uber' should be categorized as Transport Expense."
Add multiple rules for different patterns, such as keywords, merchant names, amounts, or recurring descriptions common in your bank statements.
Activating the Banking Manager
By default, the Banking Manager AI agent is active. To temporarily deactivate or reactivate it:
On the Banking Manager, locate the Action column in the Train Adam page.
Click the Play icon to activate if it is paused.
Click the Pause icon to deactivate it if it is active
This control allows you to pause custom categorization if needed, such as when reviewing or updating your rules.
How Adam Applies Rules to Transactions
Once the banking manager is activated, Adam automatically scans:
Uploaded bank statements in different file formats (PDF, Excel, CSV etc.) or synced bank transactions from connected accounts
Transactions matching your custom rules will be auto-categorized exactly as you specified.
If a transaction doesn’t match any rule, Adam uses its default models to categorize it.
This automation reduces manual data entry and ensures consistent, business-aligned classification for easier reporting and reconciliation.
💡 Tips for Effective Training
Keep instructions simple and focused.
Use specific keywords that regularly appear in your bank data.
Periodically review your rules to refine categorization as your business evolves.
Combine this training with manual reviews to ensure accuracy.