What are Tools?
Tools enable your Agent to go beyond simple conversation by performing actions such as searching the web, accessing databases, sending emails, or interacting with other systems.
When configured properly, Tools transform your Agent from a conversational interface into a powerful digital assistant that can take meaningful actions on your behalf.
Adding Tools to Your Agent
In the Agent Builder interface of Arahi AI:
-
Navigate to the “Tools” section.
-
Choose from:
-
Pre-built tool templates that address common use-cases.
-
Custom tools you’ve defined and built.
-
-
Add a tool and configure its usage settings based on your requirements.
Tool Input Configuration Options
When setting up a Tool for your Agent, you’ll choose how the Agent feeds input to the Tool. Arahi AI supports the following modes:
-
Let the Agent decide
The Agent dynamically decides the input based on context. Good for flexible workflows where you want rapid deployment. -
Set manually
You explicitly define the input fields and values for the Tool. Best suited for workflows requiring oversight or compliance controls. -
Tool output
The input to a Tool is pulled from the output of a previous Tool run (i.e., chaining Tools).
Guiding Your Agent’s Tool Usage
To ensure effective use of Tools and predictable behavior, include clear instructions in your Agent’s prompt/instruction set on when and how to use each Tool. For example:
-
When should the Agent use “Search Database” vs “Send Email”?
-
How to interpret results returned by a Tool and convert them into output or escalate.
-
What to do if a Tool fails or returns unexpected results.
Best Practices
-
Start with a few key Tools to get your Agent working reliably, then expand as your workflows grow.
-
Define roles clearly for each Tool: avoid overlapping responsibilities (e.g., two Tools doing similar things).
-
Log and monitor Tool usage and failures—this gives insight into where your Agent may need refinement or human-in-the-loop.
-
Chain Tools thoughtfully: remember that each subsequent Tool might increase latency or error risk.
-
Document your Tool inventory: keep track of descriptions, input/output schemas, dependencies, and permissions.
Example Tool Setup
Tool Name: Invoice Parser
-
Purpose: Extract line items, amounts, dates from uploaded PDF invoices.
-
Input Mode: “Set manually” — fields:
invoice_pdf_url,supplier_id. -
Chain: Output feeds into “ERP Update” Tool which uses extracted fields to update your system.
-
Agent Guidance: In the prompt, include:
“When you receive an invoice PDF, call the Invoice Parser Tool first. If the parser returns a mismatch in amounts, escalate. Otherwise, call ERP Update Tool and then send a Slack confirmation via the Slack-Notify Tool.”