Examples

Example 1 — Create an Agent

Create a New Agent

orch create agent Bob

Expected Result

Created agent: agents/bob.yaml

List Agents

orch list-agents

Run an Agent

orch run Bob --debug

Example Prompt

Summarize the latest trends in Product Marketing.

Example Agent Configuration

agent:
  id: bob
  name: Bob
  role: General Assistant

  timezone: America/Chicago
  locale: en-US

  provider:
    type: groq
    model: llama-3.3-70b-versatile

  capabilities:
    - web.request
    - datetime.local
    - memory.search
    - knowledge.search

  tools:
    - web.request
    - datetime.local
    - memory.search
    - knowledge.search

  reasoning:
    planner: true
    reflection: true

  memory:
    enabled: true

Example 2 — Create a Team

Create a Team

orch create team ContentTeam

Expected Result

Created team: teams/contentteam.yaml

List Teams

orch list-teams

Run a Team

orch run-team ContentTeam --debug

Example Team Architecture

ManagerAgent
    ↓
ResearchAgent
    ↓
WriterAgent
    ↓
ReviewerAgent
    ↓
Final Output

Example Use Case

Prompt:

Create a technical blog post explaining autonomous AI orchestration platforms.

Expected orchestration behavior:

• ResearchAgent gathers information
• WriterAgent drafts content
• ReviewerAgent validates quality
• ManagerAgent assembles final output

Example 3 — Tool Runtime

Run a Tool Directly

orch tool run datetime.local --agent Bob

Expected Output

timezone='America/Chicago' weekday='Saturday' time='08:48:28'

Example Web Request

Research the latest developments in Product Marketing automation.

The agent can dynamically invoke web.request during execution.

Example 4 — Delegation

Example Delegation Flow

ManagerAgent
  └── delegates research to ResearchAgent
  └── delegates writing to WriterAgent
  └── delegates validation to ReviewerAgent

This allows:

• Specialization
• Parallel reasoning
• Modular orchestration

Example 5 — Memory Retrieval

Search Memory

orch memory search "Product Marketing"

Agents can retrieve:

• Prior conversations
• Prior outputs
• Runtime context
• Execution history

Example 6 — Semantic Knowledge

Ingest Knowledge

orch knowledge ingest docs/

Search Knowledge

orch knowledge search "product marketing trends"

Agents can use semantic retrieval during execution.

Example 7 — Scheduled Autonomous Agent

Heartbeat Trigger

trigger:
  type: heartbeat
  interval_seconds: 3600

Example behavior:

• Run every hour
• Gather research
• Generate summary
• Store results in memory

Example 8 — Webhook-Triggered Workflow

Example flow:

• External system sends webhook
• Orchgentic trigger receives event
• Orchestration pipeline executes automatically

Example use cases:

• Ticket triage
• Incident response
• Automated reporting
• Content generation
• AI workflow automation

Example 9 — Capability Preflight

Before execution, Orchgentic validates:

• Required tools
• Provider configuration
• Team references
• Orchestration dependencies

Example Failure

ERROR: Agent requires tool 'web.request' but it is not configured.

This prevents:

• Wasted LLM calls
• Orchestration failures
• Silent runtime degradation

Example 10 — Timezone-Aware Runtime

Example Agent Time Context

timezone: America/Chicago
locale: en-US

Example Tool

orch tool run datetime.local --agent Bob

Returns:

• Localized time
• Timezone
• Weekday
• UTC offset

This enables:

• Scheduled workflows
• Region-aware orchestration
• Cloud-safe execution

Example 11 — Long-Running Research Workflow

Example Scenario

Research current Product Marketing strategies and create an executive summary.

Potential Orchestration Flow

ManagerAgent
    ↓
ResearchAgent
    ↓
Knowledge Retrieval
    ↓
WriterAgent
    ↓
ReviewerAgent
    ↓
Final Summary

Features involved:

• Web research
• Semantic retrieval
• Memory
• Delegation
• Orchestration
• Reflection

Example 12 — Autonomous Reporting System

Goal: Generate daily Product Marketing reports automatically.

Workflow:

1. Heartbeat trigger fires
2. Research agents gather data
3. Writer agent summarizes findings
4. Reviewer validates output
5. Report stored to filesystem
6. Future notification system distributes report

Example 13 — Future Workflow DAG Engine

Planned future orchestration:

START
  ↓
Research
  ↓
Branch:
  ├── Technical Summary
  ├── Executive Summary
  └── Social Content
  ↓
Review
  ↓
Publish

This will evolve into:

• Visual workflows
• Conditional routing
• Retries
• Persistent workflow state

Recommended Demo Flow

For first-time users:

orch init
orch create agent Bob
orch list-agents
orch run Bob --debug
orch tool run datetime.local --agent Bob
orch create team ContentTeam
orch list-teams
orch run-team ContentTeam --debug

This demonstrates:

• Runtime initialization
• Agent creation
• Team creation
• Tool runtime
• Provider integration
• Orchestration
• Delegation
• Team execution