Skills Overview

Large Language Models are excellent at reasoning.

They can:

Understand requests
Analyze information
Generate responses
Make decisions

But they cannot perform actions on their own.

They cannot:

Send emails
Create tickets
Query databases
Call APIs
Update CRMs
Trigger workflows

This is where Skills come in.

What Are Skills?

Skills are executable capabilities that allow Agents and AI Flows to interact with external systems and perform actions.

Think of Skills as the hands of an AI system.

User Request
      ↓
Agent Reasoning
      ↓
Skill Execution
      ↓
Business System
      ↓
Result Returned

Without Skills:

Employee:
Create a Jira ticket for this issue.

Agent:
"I cannot create tickets, but you can create one manually."

With Skills:

Employee:
Create a Jira ticket for this issue.

Agent
↓
Jira Skill
↓
Jira API
↓
Ticket Created

This transforms AI from a chatbot into an operational system.

Why Skills Matter

Enterprise AI systems need to do more than answer questions.

They need to:

Execute business actions
Update systems of record
Trigger workflows
Interact with internal tools
Automate repetitive work

Skills provide this capability.

Skills in Karyam

Skills are first-class components within Karyam and can be attached to:

Agents
AI Flows
Multi-agent systems

This allows actions to become reusable across the entire platform.

How Skills Work

When an Agent receives a request:

User Request
      ↓
Agent Reasoning
      ↓
Determine Required Action
      ↓
Execute Skill
      ↓
Receive Result
      ↓
Generate Response

Example:

Employee:
What is the status of invoice INV-1023?

Agent
↓
Invoice Lookup Skill
↓
ERP System
↓
Invoice Status Returned
↓
Response Generated

Examples of Skills

Examples include:

Communication

Send Email
Send Notifications

Data Access

Query Database
Search CRM
Lookup Employee Records

Business Operations

Create Ticket
Update Customer Record
Generate Invoice

AI Capabilities

OCR
Speech-to-Text
Text-to-Speech
Document Parsing

Skills vs MCP Tools

This distinction is important.

Capability	Skills	MCP Tools
Managed inside Karyam	Yes	No
Native platform capability	Yes	No
External protocol	No	Yes
Reusable across agents and flows	Yes	Yes
Supports third-party tool ecosystems	Limited	Yes

Think of it this way:

Skills

Native capabilities provided and managed by Karyam.

Examples:

Send Email
OCR
Database Query

MCP Tools

Capabilities exposed by external systems using the Model Context Protocol.

Examples:

GitHub MCP Server
Slack MCP Server
Jira MCP Server

Skills in AI Systems

Skills become significantly more powerful when combined with other Karyam components.

Knowledge
      +
Agent
      +
Skill
      +
Workflow
      +
Approvals
      ↓
AI System

Example:

Employee:
"My VPN access expired."

Agent
↓
Retrieve VPN Policy
↓
Check Employee Status Skill
↓
Approval Workflow
↓
Create IT Ticket Skill
↓
Notify Employee Skill

The AI system not only understands the request but completes the work.

Observability

Every skill execution is observable.

You can monitor:

Inputs
Outputs
Execution duration
Failures
Retry attempts

This allows teams to debug and optimize production AI systems.

Security and Governance

Skills operate within organizational boundaries.

Karyam supports:

Role-Based Access Control (RBAC)
Permissions
Audit Logs
Approval Workflows
Human-in-the-Loop (HITL)

This ensures AI actions remain secure and governed.

The Karyam Philosophy

Traditional AI systems stop at understanding.

Karyam goes further.

Understand
    ↓
Reason
    ↓
Act

Skills are what enable AI systems to move from answering questions to performing work.

Next Steps

Now that you understand what Skills are, continue with:

Built-In Skills
Request A Skill

Together these allow organizations to connect AI systems with the tools and processes that power their business.