Agents

Agents are intelligent orchestrators that manage multiple playbooks simultaneously, enabling complex multi-step workflows that span different tools and data sources.

Think of it this way:

Playbook = Single automation task (e.g., “fetch weather”)
Agent = Coordinator of multiple tasks (e.g., “morning routine: fetch weather, check calendar, send daily brief”)

What is an Agent?

An Agent is a smart container that:

🎯 Orchestrates Multiple Playbooks - Run several automations as one cohesive workflow
⏰ Scheduled Execution - Run all playbooks at specified times or intervals
🔄 Sequential or Parallel - Control how playbooks execute (one after another or simultaneously)
📊 Unified Monitoring - Track all playbook executions in one place
🤖 NIRA-Powered - Build complex agents through conversation

Agent vs Playbook

Playbook

Single automation task

One specific workflow
Uses one or more MCP tools
Runs independently
Example: “Send Slack notification”

Agent

Multi-playbook orchestrator

Coordinates multiple playbooks
Manages complex workflows
Scheduled or interval-based
Example: “Daily morning routine”

When to Use Agents

Use Agents when you need to:

Coordinate Multiple Data Sources

Example: Pull data from Salesforce, enrich with market data from Finnhub, analyze with AI, and post results to SlackWhy Agent: Multiple playbooks working together, each handling one data source

Execute Sequential Workflows

Example: Morning routine - check weather → check calendar → check emails → generate daily brief → send to SlackWhy Agent: Playbooks need to run in specific order, passing data between steps

Recurring Multi-Step Processes

Example: Hourly: monitor system health → check API quotas → update dashboard → alert if issuesWhy Agent: Complex workflow that repeats on a schedule

Conditional Workflow Branching

Example: Sales pipeline: check new leads → qualify leads → assign to reps → notify team, but only for qualified leadsWhy Agent: Logic spans multiple playbooks with conditional execution

Creating an Agent with NIRA

The easiest way to create an agent is to ask NIRA:

Describe Your Multi-Step Workflow

Tell NIRA about the complex automation you want to build

You: “I need a morning routine agent that checks the weather, pulls my calendar for today, summarizes my emails, and sends me a daily brief in Slack every weekday at 7am”

NIRA Analyzes & Proposes Structure

NIRA identifies this needs multiple playbooks and suggests an agent

NIRA says: 🤖“Perfect! This is a great use case for an Agent. I’ll create a ‘Morning Routine Agent’ with 4 playbooks:

Weather Playbook - Fetches forecast for your location
Calendar Playbook - Gets today’s events from Google Calendar
Email Playbook - Summarizes unread emails
Brief Playbook - Compiles everything and sends to Slack

These will run sequentially so each step completes before the next starts.Schedule: Weekdays at 7:00 AMSound good?”

Configure API Keys

Provide credentials for each service the agent will use

Review Agent Configuration

See all playbooks, their execution order, and schedule

Deploy & Monitor

Agent goes live and you can track each playbook execution

Agent Configuration

Scheduling Options

Agents can be configured with flexible scheduling:

Time-Based
Interval-Based
Event-Triggered

Run at specific times:

Daily at 9:00 AM
Weekdays at 8:00 AM and 5:00 PM
Every Monday at 10:00 AM
First day of month at 9:00 AM

{
  "schedule": {
    "type": "time",
    "time": "09:00",
    "days": ["monday", "tuesday", "wednesday", "thursday", "friday"],
    "timezone": "America/Panama"
  }
}

Execution Modes

Control how playbooks within an agent execute:

Sequential

Playbooks run one after anotherBest for:

Dependent steps
Data flows between playbooks
Resource-intensive tasks

Example: Fetch data → Process → Analyze → Report

Parallel

Playbooks run simultaneouslyBest for:

Independent tasks
Speed optimization
Multiple data sources

Example: Check 5 APIs at once, compile results

Playbook Dependencies

Define relationships between playbooks:

{
  "playbooks": [
    {
      "id": "fetch-sales-data",
      "name": "Fetch Sales Data",
      "dependencies": []
    },
    {
      "id": "fetch-market-data",
      "name": "Fetch Market Data",
      "dependencies": []
    },
    {
      "id": "analyze-data",
      "name": "Analyze Combined Data",
      "dependencies": ["fetch-sales-data", "fetch-market-data"]
    },
    {
      "id": "send-report",
      "name": "Send Report to Team",
      "dependencies": ["analyze-data"]
    }
  ]
}

Real-World Agent Examples

1. Sales Intelligence Agent

Sales Intelligence Agent

Goal: Provide daily competitive intelligence to sales teamPlaybooks:

Competitor Monitor - Scrape competitor pricing pages
Market Data - Fetch relevant market trends (Finnhub)
CRM Sync - Pull today’s opportunities (Salesforce)
AI Analysis - Analyze data and generate insights
Report Generator - Create formatted report
Slack Notifier - Send to #sales channel

Schedule: Every weekday at 6:00 AM Execution: Sequential (each step needs previous data)

2. System Health Agent

System Health Monitor

Goal: Monitor application health and alert on issuesPlaybooks:

API Health Check - Test all API endpoints
Database Monitor - Check database performance
Log Analyzer - Scan for errors in last hour
Quota Monitor - Check API usage limits
Alert Generator - Create alerts if issues found
PagerDuty Trigger - Escalate critical issues

Schedule: Every 15 minutes Execution: Parallel (all checks run simultaneously)

3. Content Distribution Agent

Content Distribution Agent

Goal: Automatically publish content across multiple platformsPlaybooks:

Content Fetch - Get approved posts from Notion
Image Optimizer - Resize/optimize images
Twitter Publisher - Post to Twitter
LinkedIn Publisher - Post to LinkedIn
Slack Notifier - Notify team of publications
Analytics Tracker - Log publication to database

Schedule: Daily at 9:00 AM and 3:00 PM Execution: Parallel for social posts, sequential for tracking

Managing Agents

Agent Dashboard

View all your agents in one place:

📊 Execution history and success rates
⏱️ Next scheduled run times
🎯 Active vs paused agents
📈 Performance metrics per playbook

Editing Agents

Modify agents at any time:

➕ Add new playbooks
➖ Remove playbooks
🔄 Change execution order
⏰ Update schedule
🔧 Adjust configurations

Pro Tip: Ask NIRA to modify your agent: “Add a playbook to the morning routine that checks my stock portfolio”

Pausing & Resuming

Temporarily disable agents without deleting:

Pause before holidays or maintenance
Resume when ready
Edit while paused
View historical data even when paused

Advanced Features

Conditional Execution

Run playbooks based on conditions:

Only send alerts if threshold exceeded
Skip steps if data is empty
Branch based on time of day or day of week

{
  "playbook": "send-alert",
  "condition": "temperature > 100"
}

Data Passing Between Playbooks

Share data between playbooks in an agent:

Output from one playbook becomes input to next
Access shared agent context
Transform data between steps

{
  "playbook": "analyze-data",
  "inputs": {
    "salesData": "${fetch-sales-data.output}",
    "marketData": "${fetch-market-data.output}"
  }
}

Error Handling & Retries

Configure resilient agents:

Retry failed playbooks automatically
Continue on error vs stop agent
Fallback playbooks
Error notifications

{
  "errorHandling": {
    "retry": true,
    "maxRetries": 3,
    "retryDelay": "5m",
    "onError": "continue"
  }
}

Agent Variables

Define reusable variables across playbooks:

Environment configurations
Shared settings
Dynamic values

{
  "variables": {
    "slackChannel": "#sales-team",
    "reportFormat": "detailed",
    "timezone": "America/Panama"
  }
}

Monitoring & Logs

Track agent performance in detail:

Execution History

See every agent run with timestamp
Playbook-by-playbook status
Execution duration
Success/failure indicators

Detailed Logs

Each playbook’s output
MCP tool interactions
Error messages and stack traces
Data transformations

Performance Metrics

Average execution time
Success rate percentage
Resource usage
Cost tracking (API calls)

Best Practices

Start Simple

Begin with 2-3 playbooks and add more as needed

Name Clearly

Use descriptive names: “Morning Routine” not “Agent 1”

Test Individually

Test each playbook before combining into agent

Monitor Initially

Watch first few runs closely to catch issues

Handle Failures

Configure retries and error handling upfront

Document Purpose

Add descriptions so team understands agent’s goal

Troubleshooting

Agent didn't run on schedule

Check:

Agent is not paused
Schedule timezone is correct
Schedule cron expression is valid
System status page for outages

One playbook failed, stopped entire agent

Solution:

Configure error handling to “continue” instead of “stop”
Add retry logic to the failing playbook
Review playbook logs to fix underlying issue

Playbooks running in wrong order

Check:

Execution mode (sequential vs parallel)
Playbook dependencies are configured correctly
Review execution logs for actual order

Agent runs too slowly

Optimize:

Switch independent playbooks to parallel execution
Reduce API call frequency in playbooks
Cache frequently-accessed data
Review MCP tool performance

What’s Next?

Build Your First Agent

Ask NIRA to create a multi-playbook workflow

Learn About Playbooks

Understand the building blocks of agents

Explore MCP Tools

See what tools your agent can use

Activity Logs

Track your agent’s execution and success

Get Started

Core Features

Developer Tools

​Agents

​What is an Agent?

​Agent vs Playbook

Playbook

Agent

​When to Use Agents

​Creating an Agent with NIRA

​Agent Configuration

​Scheduling Options

​Execution Modes

Sequential

Parallel

​Playbook Dependencies

​Real-World Agent Examples

​1. Sales Intelligence Agent

Sales Intelligence Agent

​2. System Health Agent

System Health Monitor

​3. Content Distribution Agent

Content Distribution Agent

​Managing Agents

​Agent Dashboard

​Editing Agents

​Pausing & Resuming

​Advanced Features

​Monitoring & Logs

​Execution History

​Detailed Logs

​Performance Metrics

​Best Practices

Start Simple

Name Clearly

Test Individually

Monitor Initially

Handle Failures

Document Purpose

​Troubleshooting

​What’s Next?

Build Your First Agent

Learn About Playbooks

Explore MCP Tools

Activity Logs

Agents

What is an Agent?

Agent vs Playbook

When to Use Agents

Creating an Agent with NIRA

Agent Configuration

Scheduling Options

Execution Modes

Playbook Dependencies

Real-World Agent Examples

1. Sales Intelligence Agent

2. System Health Agent

3. Content Distribution Agent

Managing Agents

Agent Dashboard

Editing Agents

Pausing & Resuming

Advanced Features

Monitoring & Logs

Execution History

Detailed Logs

Performance Metrics

Best Practices

Troubleshooting

What’s Next?