Workflow Builder

Design a workflow: agents, sub-agents, approvals, escalation, tools.

The system does not use a fixed agent sequence. The workflow is dynamically selected based on task type, required skills, risk, tools, confidence score, and approval rules.
Core design principle: use the minimum required agents for a task. Do not involve every agent unless the workflow actually needs them.

1. Workflow basics

Workflow name
Use case
Starting agent
Lead agent

2. Agents

Agents involved
Optional agents
Agents excluded

3. Steps & gates

Sequential steps
Parallel steps
Approval gates
Escalation points
Human review points

4. Sub-agents & handoffs

Sub-agents under each main agent
Available sub-agents
• Pre-Sales Agent → Software Pre-Sales Sub-Agent
• Pre-Sales Agent → Cisco Pre-Sales Sub-Agent
• Pre-Sales Agent → Cloud Pre-Sales Sub-Agent
• Pre-Sales Agent → AI / Data Pre-Sales Sub-Agent
• Pre-Sales Agent → Security Pre-Sales Sub-Agent
• IT-Security Agent → Identity & Access Sub-Agent
• IT-Security Agent → Network Security Sub-Agent
• IT-Security Agent → Cloud Security Sub-Agent
• IT-Security Agent → Data Privacy Sub-Agent
• IT-Security Agent → Compliance Sub-Agent
• IT-Security Agent → Vulnerability Review Sub-Agent
• IT-Security Agent → SOC / SIEM Sub-Agent
• Legal Agent → Contract Review Sub-Agent
• Legal Agent → Liability Clause Sub-Agent
• Legal Agent → Data Protection Clause Sub-Agent
• Legal Agent → Compliance Clause Sub-Agent
• Legal Agent → Deviation Detection Sub-Agent
• Legal Agent → Redline Suggestion Sub-Agent
• Cloud-Ops Agent → AWS Ops Sub-Agent
• Cloud-Ops Agent → Azure Ops Sub-Agent
• Cloud-Ops Agent → GCP Ops Sub-Agent
• Cloud-Ops Agent → FinOps Sub-Agent
• Cloud-Ops Agent → SRE / Reliability Sub-Agent
• Cloud-Ops Agent → Backup & DR Sub-Agent
• Cloud-Ops Agent → Monitoring & Observability Sub-Agent
• Finance Agent → Pricing Validation Sub-Agent
• Finance Agent → Margin Check Sub-Agent
• Finance Agent → Discount Approval Sub-Agent
• Finance Agent → Payment Terms Sub-Agent
• Finance Agent → Revenue Forecast Sub-Agent
• Product Agent → Product Fitment Sub-Agent
• Product Agent → Feature Validation Sub-Agent
• Product Agent → Roadmap Check Sub-Agent
• Product Agent → Product Limitation Sub-Agent
• Operations Agent → Delivery Planning Sub-Agent
• Operations Agent → Resource Planning Sub-Agent
• Operations Agent → SLA Review Sub-Agent
• Operations Agent → Risk & Dependency Sub-Agent
Handoff rules between agents

5. Completion & thresholds

Completion criteria
Risk thresholds
Confidence threshold
Stop conditions
Escalation conditions

6. Tools available per workflow

Tools

Conditional routing rules applied

  • If task type = security review → Route to IT-Security Agent
  • If task type = contract review → Route to Legal Agent
  • If task type = cloud cost review → Route to Cloud-Ops Agent + Finance Agent
  • If task type = proposal creation → Route to Pre-Sales Agent
  • If task mentions discount or margin → Include Finance Agent
  • If task mentions contract, liability, NDA, or legal terms → Include Legal Agent
  • If task mentions data privacy, access control, vulnerability, SOC, or compliance → Include IT-Security Agent
  • If task mentions delivery timeline, resources, implementation, or SLA → Include Operations Agent
  • If confidence below threshold → Escalate to Coordinator or Human Reviewer
  • If risk = high → Escalate to MD or relevant human approver
  • If task can be completed by one agent → Do not involve unnecessary agents

Generated workflow preview

Draft
{
  "name": "Custom Cloud Security Review",
  "useCase": "Joint cloud cost + security posture review for a strategic customer.",
  "starting": "Coordinator Agent",
  "lead": "IT-Security Agent",
  "involved": [
    "Coordinator Agent",
    "IT-Security Agent",
    "Operations Agent"
  ],
  "optional": [
    "Finance Agent",
    "Legal Agent"
  ],
  "excluded": [
    "Sales Agent",
    "HR Agent",
    "Product Agent"
  ],
  "sequential": "Coordinator → IT-Security → Operations → Coordinator",
  "parallel": "IT-Security ∥ Operations during assessment phase",
  "approvalGates": "Human CISO sign-off; MD approval if risk = High",
  "escalation": "Risk High → MD; confidence < 0.7 → Coordinator",
  "humanReview": "Final report → Human Security Reviewer",
  "subAgentNotes": "IT-Security: IAM, Cloud Security, Compliance Sub-Agents",
  "handoffs": "IT-Security passes risk score → Operations adjusts plan",
  "completion": "Risk report + remediation plan delivered & approved",
  "riskThreshold": "High = escalate, Medium = note, Low = auto-pass",
  "confidenceThreshold": "0.75",
  "stop": "Customer pulls request OR scope changes",
  "escalationCond": "Critical CVE OR data residency violation",
  "tools": [
    "Security policy RAG",
    "Risk scanner",
    "Cloud cost analyzer"
  ]
}