Skip to main content

Build Your First Agent

In this quickstart guide, you will learn to Create a new Agent.

Agent Development Simplified​

  • Build AI Agents with No-code
  • Customize & Extend to complete your use case.
  • Use AI Agents in workflows or as chat.
info

Building Steps and Resources​

We will be building CRM AI Agents with context from CRM database and sales documents.

  • Build AI Agents and AI Workflows with pattern chaining.
  • Make AI-Ready Data available.
  • Optionally Provision RAG Context (Sales Documents).
  • Use Tools and Resources
    • EasyManage On-Premise Edition
    • Your CRM database accessible via JDBC
    • For some quick assistance, AI Tool like Cursor or Copilot in VS Code
    • Model - LLM Provider and API Key
    • On-premise host: Win 11, RAM 16GB, HD 256GB.
      • With Docker Desktop, Java 17 or higher.

Setup EasyManage On-Premise​

AI-Ready Data​

Make AI-Ready Data available with EasyManage building.

Process Tables and Views​

  • Steps in EasyManage Builder Studio:
    • Sign-Up/Use EasyManage Id: CRMApp01
    • Setup Db Source
    • Import Tables
    • Download Table Schema, e.g. TableSchemaDnld_CRMApp01.sql
info

For Process Tables and Views, Use any AI Tools Editors like:

  • VS Code with Copilot chat
  • Cursor

Use AI Tools to generate final SQL script.​

  • Giving as context TableSchemaDnld_CRMApp01.sql, Ask AI model
  • Sensitive data columns filtering:
Study columns and mark columns , put comment on right side, for columns that come under:
PII (Personally Identifiable Information)
Sensitive Personal Information
Sensitive Data
Confidential Business Information
  • Create singular views on each table:
create new sql script with view creation for each table, 
selecting all columns, 
except those annotated for sensitive data with '-- [ ]', 
also add '_evw' suffix to view name
  • In views script, Rename, if have, key column as ID. Change: ID, To: ID keyId,
  • See example below for one final table view script from generated views script e.g. TableSchemaDnld_CRMApp01_Views.sql:
/* View for table: Accounts */
CREATE VIEW Accounts_evw AS
SELECT
  ID keyId,
  AccountName,
  AccountDescription,
  ...
FROM Accounts;

...
  • Load/Create these views in CRM database: using views script above.

Use AI Tools to Analyze table schemas for AI workflows​

  • Giving as context TableSchemaDnld_CRMApp01.sql, Ask AI model
  • Ask AI What AI Workflows Can Implement:
study all table schema and suggest what AI workflows can be implemented based on tables
  • Save response, Study it and short-list first AI Agent workflow you want to build.

Build AI Agents​

  • No-code Build AI Agents and AI Workflows with pattern chaining.

Build Agent MCP Packs​

  • Steps in EasyManage Builder Studio:
    • Sign-Up/Use EasyManage Id: CRMApp
    • Setup Db Source
    • Import Tables: Import new views only, using search qualifier '_evw'
    • Configure Project
    • Select Tables Imported xxxx folder for build selection
    • Generate Code
    • Download
  • Next, follow steps in Chapter 10 : Agent MCP Packs

Agents & Dashboard​

Use AI Chat​

  • Conversational AI
    • Using AI Chat allows users to ask conversational questions about their data and get analytics.
  • Use AI Chat, See Agent Chat

agent-chat-view-chart.png

Run Generated Agents​

Run Generated No-Code Agents.

  • AI Agent Workflows as Agent Projects
    • Use sample generated Agents, customize to your needs.
    • See Agent Project

agent-proj.png

agent-test.png

tip

Generated No-Code Agent Projects help to jumpstart building your specific Agentic Workflows.

Build Your 1st Agent​

Once familiar with How Agent Project is organized and works, you are ready to build AI Agent with No-Code.

For example during case study Section 5-3 : Case Study SuiteCRM, Para: Ask AI What AI Workflows Can Implement, below Agent Workflow is suggested.

Agent: Cross-sell / upsell recommendations

Uses: AosProductsQuotesVw joined by groupId/parentId to quotes/opps, with account segments from AccountsVw.
Outcome: Product bundling suggestions per account/opportunity.
  • Similarly, Study your earlier short-listed AI Agent workflow description, and how it should be built.
  • Implement the steps in workflow as AI Tasks i.e. Agent Tasks.
  • Run and refine your Agent to get desired response.
info

Congratulations, you have built AI Agent.

Checklist For AI Agents Building​

Reference Checklist & Steps For AI Agents Building to production roll-out.

See Credentials Setup for production use.

EasyManage On-Premise​

  • Setup EasyManage On-Premise Edition
    • Import tables
    • Table Schema | Download
    • Table View SQL | Download

Model - LLM & AI Tool​

  • Provision/Setup
    • Openrouter.ai API Key with Credits
    • AI Tool / Editor, setup any tool for GenAI usage, e.g.:
      • VS Code with Copilot
      • Chatgpt

AI-Ready Data​

  • Process Tables and Views
    • Analyze table schemas for PII, Sensitive data columns filtering
    • Get Processed Tables and Views
  • In EasyManage Builder
    • Import Processed Table Views
    • Project Configure → Project Settings - Security Configuration
      • Select Your Security Method
    • Generate Code, Download
    • Build Agent MCP Packs, Deploy
    • Configure Agent Server
  • AI-Ready data available via MCP Tools

No-Code Agents​

  • In EasyManage Builder, Import Tables Action in last step, also creates No-Code Agents
  • Agents & Dashboard
    • Build and test generated no-code AI Agents

AI Chat​

  • AI Chat
    • Use Agent Chat (Conversational AI)

Production AI Agents​

  • Analyze table schemas for AI workflows
    • In AI Tool
      • Ask AI What Top 20/30 AI Agents And Workflows Can Implement
    • Short-list Phase 1 AI Agents and Workflows you want to build
  • Select Phase 1 AI Agent To Implement
    • e.g. "Input Items Procurement Optimization Engine"
    • In AI Tool
    • Ask AI: "To build AI Agent <Agent Name>, Study the table schema, and give <your database type & ver> queries for building agent"
  • Verify, validate and finalize database queries
    • Run each query against your database and verify
    • Build intermediate Views and then get final Agent Query View.
  • AI Agent Build and Roll-out
    • Customize No-Code Agents to complete Use Case
    • Or Create New Agent Project
      • Add all your workflow tasks or queries
      • Add final Agent Query Task
    • Emd User Acceptance
      • End Users run, test Agents and validate
      • As required, Make changes to workflows and complete acceptance testing
    • Production Roll-out
      • Roll-out Agent to Production.

Scheduling & Integrations​

As per requirement,

  • Schedule AI Agent Runs using Kestra
  • Integrate AI Agents in your Frontend/App with available REST interface

Credentials Setup​

Set strong passwords and proper credentials for Production use.

caution

Do NOT use default passwords for production setup.

  • EasyManage On-Premise edition
    • File .env
      • MySQL credentials: MYSQL_ROOT_PASSWORD, EMDB_PASSWORD
    • File em_dirs\warextracted\em\WEB-INF\classes\em_param1_en_US.properties
      • em.db.mysql.password : same value as EMDB_PASSWORD
    • File docker-compose.yml
      • em_app Tomcat image: Use this image for Agent Svr Security Basic Auth
      • image: tomcat:9.0-jdk8
  • EasyManage Builder account passwords
    • When you Sign-Up for below User Accounts (or your custom accounts)
      • ADMIN, ApplicationEm, emDeveloper, ...
    • Project Configure : Project Settings - Security Configuration
      • Passwords for emUser and emAdmin
  • Gnerated code Agent MCP Pack Auth
    • Credentials for Db Sources
    • Basic Auth credentials