Building Your First AI Agent: Step-by-Step Guide with Craveva AI Enterprise

Founder Summary

• Audience: CXOs and founders running catering, franchise groups, casual dining, cloud kitchens.
• Core outcomes (what moves the business):
• Operational consistency: standardize execution across outlets using Craveva AI Enterprise agents + data layer.
• Cost savings: reduce waste and procurement errors, automate purchasing cycles with Craveva AI Enterprise.
• Sales lift: increase AOV and conversion with Craveva AI Enterprise sales agents on web/WhatsApp/kiosks.
• Time savings: remove manual exports, reporting, and SOP Q&A with Craveva AI Enterprise automation.

Platform flow (high level)

• Deployment layer: deploy agents to WhatsApp, web widget, kiosks, or internal tools with Craveva AI Enterprise.
• AI layer: agents query and act on governed data (no fragile spreadsheet workflows) in Craveva AI Enterprise.
• Data layer: connect POS, databases, Google Drive, and APIs into a unified view inside Craveva AI Enterprise.

Operating Model (how teams run it)

• Ops defines workflows (ordering, inventory alerts, SOP answers, customer responses) in Craveva AI Enterprise.
• Finance sets guardrails (approval thresholds, budgets, audit trail) in Craveva AI Enterprise.
• IT connects data sources once; rollout scales outlet-by-outlet via Craveva AI Enterprise multi-outlet deployment.
• Leadership tracks KPI movement weekly and expands successful automations with Craveva AI Enterprise.

Setup (30–60 minutes to first value)

• Deploy to the workflow: WhatsApp/web/kiosk/internal portal using Craveva AI Enterprise.
• Measure ROI and operational impact, then replicate across brands/outlets with Craveva AI Enterprise.
• Connect data sources (POS + databases + Drive + APIs) in Craveva AI Enterprise.
• Start with 2–3 agents: Procurement (cost), Sales (revenue), Analytics (visibility) in Craveva AI Enterprise.

CXO KPIs

• PO approval turnaround and exception rate
• Peak-hour conversion vs queue time
• Training completion rate and knowledge check scores
• Ingredient substitution rate and margin impact
• Expedite frequency and cost (urgent orders)

Explore the Platform

• Architecture: /solutions/architecture
• Deployment: /solutions/deployment
• Documentation: /documentation
• Models: /ai-models
• Templates: /templates

Introduction

Building your first AI agent with Craveva AI Enterprise is straightforward thanks to the AI-Assisted Agent Builder, which automatically detects entity relationships, maps data, and generates prompts. This guide walks you through creating a customer service agent step-by-step, but the same process applies to any agent type.

Prerequisites

Before starting:

• Craveva AI Enterprise account (sign up at platform)
• Data source ready (optional but recommended): POS system, database, or Google Drive
• Clear use case: Know what you want the agent to do
• 5-10 minutes: First agent takes just minutes to build

Step 1: Define Your Use Case

Start with a clear objective. For this example, we'll build a Customer Service Agent that:

• Answers customer questions about menu items
• Tracks orders
• Handles common FAQs
• Processes refund requests

Success Metrics:

• Response time < 2 seconds
• 90%+ accuracy on common questions
• Customer satisfaction > 4.5/5

Step 2: Connect Data Sources

Before building the agent, connect your data sources:

1. Navigate to Data Sources: Click "Data Sources" in dashboard
2. Add Data Source: Click "Add Data Source"
3. Select Type: Choose from:
   • POS Systems: Qashier, Eats365, Raptor, Micros, Toast, Lightspeed, StoreHub, MEGAPOS
   • Databases: 12 types supported - PostgreSQL, MySQL, MongoDB, BigQuery, Snowflake, Redshift, Athena, ClickHouse, Trino, SQL Server, Oracle, DuckDB
   • APIs: REST API, GraphQL API
   • Google Drive: Connect Google Drive/Google Docs
   • Files: Upload CSV, Excel, PDF, JSON files
4. Enter Credentials: Provide connection details
5. Test Connection: Platform validates automatically
6. Save: Connection saved for use by agents

For Customer Service Agent: Connect your POS system (Qashier) to access menu items, prices, and order history.

Time: 5-10 minutes

Step 3: Create Agent with AI-Assisted Builder

1. Navigate to Agent Builder: Click "Create Agent" in dashboard
2. Select Template (Optional): Choose "AI Customer Service Agent" template, or start from scratch
3. Enter Agent Details:
   • Name: "Customer Service Agent"
   • Description: "Handles customer inquiries, order tracking, and FAQs"
   • Category: Customer Service
4. Enable AI Mode: Toggle "AI-Assisted Builder" ON
5. Select Data Sources: Choose your connected data sources (Qashier POS)
6. Click "Create": AI-Assisted Builder starts automatic configuration

Step 4: AI-Assisted Automatic Configuration

The AI-Assisted Agent Builder automatically:

1. Detects Entity Relationships: Analyzes your data to find products, orders, customers, etc.
2. Maps Data Entities: Maps product names, prices, order IDs to standard formats
3. Generates Prompts: Creates system prompts for customer service tasks
4. Creates MDL: Generates Modeling Definition Language for data access
5. Configures Capabilities: Sets up order tracking, FAQ answering, refund processing

Time: 1-2 minutes (automatic)

Step 5: Configure AI Model

Craveva AI Enterprise provides 342+ AI models via Craveva LLM Router:

1. Model Selection: Choose from dropdown:
   • Recommended: Claude 3 Sonnet (excellent for customer service)
   • Cost-Effective: Claude 3 Haiku (for high volume)
   • Premium: GPT-4 Turbo (for complex queries)
2. Auto-Routing (Optional): Enable automatic model routing based on task complexity
3. Custom Configuration: Override auto-routing if needed

For Customer Service: Claude 3 Sonnet recommended for best dialogue quality.

Step 6: Test with Live Preview

Craveva AI Enterprise provides Live Preview for testing:

1. Open Live Preview: Click "Test Agent" button
2. Try Sample Queries:
   • "What's on the menu today?"
   • "Track my order #12345"
   • "What are your opening hours?"
   • "I want a refund for order #12345"
3. Review Responses: Check accuracy and relevance
4. Iterate: Adjust prompts or data mappings if needed

Time: 5-10 minutes of testing

Step 7: Deploy Agent

Deploy your agent to any platform:

Option 1: WhatsApp Business API

1. Navigate to Deployments: Click "Deployments" section
2. Select WhatsApp: Choose WhatsApp Business
3. Enter Credentials:
   • Business Account ID
   • API Access Token
   • Phone Number ID
   • Webhook Verify Token
4. Test Connection: Verify credentials
5. Deploy: One-click activation

Time: 5 minutes

Option 2: Website Widget

1. Navigate to Deployments: Click "Deployments"
2. Select Website Widget: Choose widget deployment
3. Copy Code: Platform generates code in 6 formats:
   • JavaScript
   • TypeScript
   • React
   • Vue
   • Svelte
   • Angular
4. Embed in Website: Paste code into your website
5. Test: Agent appears as chat widget

Time: 2 minutes

Option 3: Custom Integration

1. Get API Endpoint: Platform provides agent API endpoint
2. Integrate: Use endpoint in your custom system
3. Test: Verify integration works

Step 8: Monitor and Optimize

Monitor agent performance:

1. Analytics Dashboard: View real-time metrics:
   • Usage: Number of queries, response times
   • Performance: Accuracy, customer satisfaction
   • Costs: AI model usage costs
   • Errors: Failed queries, issues
2. User Feedback: Collect feedback from customers
3. Optimize: Adjust prompts, data mappings, or model selection based on data
4. Iterate: Continuously improve based on performance

Complete Example: Customer Service Agent

Setup (15 minutes):

1. Connected Qashier POS system (5 min)
2. Created agent with AI-Assisted Builder (2 min)
3. Selected Claude 3 Sonnet model (1 min)
4. Tested with Live Preview (5 min)
5. Deployed to WhatsApp (2 min)

Results (after 1 month):

• 500+ queries/day handled automatically
• < 2 second average response time
• 92% accuracy on common questions
• 4.6/5 customer satisfaction
• $336K annual savings (10 outlets) from reduced staff time

Best Practices

1. Start Simple: Begin with basic capabilities, add features gradually
2. Test Thoroughly: Use Live Preview to test before deployment
3. Monitor Closely: Check analytics daily for first week
4. Iterate Based on Data: Use performance data to improve
5. Use Templates: Start with templates, customize as needed

Common Mistakes to Avoid

1. Over-Complicating: Start with simple use case, expand later
2. Insufficient Testing: Test thoroughly before going live
3. Poor Data Quality: Ensure data sources are accurate and up-to-date
4. Ignoring Analytics: Monitor performance and optimize continuously
5. Wrong Model Selection: Use recommended models for your use case

Next Steps

Once your first agent is working:

1. Build More Agents: Create AI Sales Agent, AI Procurement Assistant, etc.
2. Connect More Data: Add databases, APIs, Google Drive
3. Deploy to More Platforms: Add Messenger, Telegram, websites
4. Scale: Deploy to multiple outlets
5. Optimize: Continuously improve based on data

Conclusion

Craveva AI Enterprise makes building AI agents accessible to everyone. The AI-Assisted Agent Builder automatically handles data mapping, prompt generation, and configuration, reducing setup time from hours to minutes. With Live Preview for testing, one-click deployment to any platform, and comprehensive analytics, you can build, test, and deploy your first agent in under 20 minutes. Start with a simple use case, test thoroughly, and iterate based on performance data to create powerful AI agents that transform your F&B operations.

KPIs to track