Quick Start Guide

Welcome to building your first Agent with Arahi AI

Welcome to the fastest route to launching your first AI Agent on Arahi AI. Whether you’re new to agent-based automation or you’re looking to deploy your next business process, this guide will walk you through creating your first Agent in just a few minutes.

Three ways to begin

Whichever path you take, you’ll still customise your Agent’s prompt, tools, and settings so it fits your workflow perfectly.

  1. Create an Agent with a Prompt
    Use our “Creator” Agent: just describe what you need and Arahi AI will generate a working draft Agent (prompt template + tool placeholders) so you can get started quickly.
    Tip: If speed is your goal, this is the fastest route.

  2. Clone an Agent Template
    Choose from a library of pre-built Agent templates developed for common business use-cases (e.g., lead qualification, invoice processing, customer-support triage). Clone and tweak to match your exact process.

  3. Build an Agent from Scratch
    If you have a unique workflow or advanced needs, start from a blank slate. This guide will help you through each building block—prompts, tools, triggers, settings—so you can craft a fully tailored Agent.

Getting started step-by-step

Below is a streamlined walkthrough to get your Agent live.

Step 1: Create Your Agent

  • Navigate to the Agents section in Arahi AI.

  • Click Create New Agent.

  • Enter: Agent name, description, avatar/icon (optional) to give it identity.

  • Select whether you’ll use a template, use Agent Invent, or start from scratch.

Step 2: Define Prompt & Instructions

  • Craft or refine the prompt that will power the Agent’s behaviour.

  • Specify system instructions (tone, context), user input format, expected output.

  • Example: “You are InvoiceBot, you receive an invoice PDF and you extract line-items, match to PO, flag mismatches and update database.”

  • Set boundary conditions and escalation logic (when to hand off to a human).

Step 3: Add Tools & Integrations

  • Under “Tools / Connected Resources”, click + Add Tool.

  • Choose from built-in tools (e.g., document parser, vector-search, CRM integration, email sender) or build a custom tool.

  • For each tool: assign name, description, trigger conditions, input/output schema.

  • Example: attach a document-parser tool to handle PDF uploads; attach a webhook tool to send updates to your CRM.

Step 4: Set Triggers & Workflow Logic

  • Define what triggers the Agent: e.g., uploaded document, Slack message, webhook event, scheduled job.

  • Configure branching logic: e.g., if amount > $10 k, escalate; else auto-approve.

  • Optional: assemble multi-step workflows if needed (Agent does Task A → triggers Tool B → reviews → sends report).

Step 5: Test & Validate

  • Run a few test inputs to see how the Agent behaves.

  • Check that tools execute correctly and the outputs make sense.

  • Validate edge cases, error handling, escalation paths.

  • Adjust prompt, tool configuration, or logic based on results.

Step 6: Deploy & Monitor

  • When you’re happy with test runs, switch Agent from “Draft” to “Active”.

  • Assign access permissions (which users or teams can trigger this Agent).

  • Monitor key metrics: number of tasks processed, error rate, time saved, human escalations.

  • Set up alerts for anomalies (e.g., spike in failed tasks).

  • Regularly review performance and refine: update prompts, add new tools, adjust logic.

Tips & Best Practices

  • Start small, iterate fast: Launch a minimal viable Agent (MVA) to solve one pain point, then expand.

  • Define success metrics: e.g., “reduce manual invoice review time by 50%” — track this.

  • Use human-in-the-loop when needed: For edge-cases, degrade gracefully to a human operator rather than break.

  • Monitor and refine: Agent behaviour will drift or edge-cases will appear — plan for iteration.

  • Brand the Agent: Give it a clear name and identity so team members know “who” they’re interacting with.

  • Secure and compliant: Since Arahi AI handles business documents and user data (including personal data/IP), ensure your tools and workflows comply with your GDPR, SOC 2 and other compliance regimes.

That’s it you’re on your way to launching your first Agent with Arahi AI.
If you hit a wall, check our support docs or reach out to support at support@arahi.ai.

Updated on
Knowledge Base
Build AI Agent