Business process automation has evolved far beyond simple if-then rules. With OpenClaw on Tencent Cloud Lighthouse, you can build AI-powered process automation that handles complex, judgment-heavy workflows — the kind that traditionally required human decision-making at every step.
Traditional RPA (Robotic Process Automation) tools excel at repetitive, rule-based tasks. But they break when:
OpenClaw bridges this gap with AI-powered decision nodes embedded directly in your automation workflows.
┌──────────────────┐ ┌──────────────────┐ ┌──────────────┐
│ Trigger Layer │ │ OpenClaw Core │ │ Action │
│ - Webhooks │────▶│ + Process │────▶│ Layer │
│ - Schedules │ │ Skills │ │ - APIs │
│ - Events │ │ + AI Engine │ │ - Databases │
│ - File Drops │ │ │ │ - Services │
└──────────────────┘ └──────────────────┘ └──────────────┘
│
┌──────────────┐
│ Lighthouse │
└──────────────┘
A 2-core 4GB Lighthouse instance handles most process automation workloads. Scale to 4-core 8GB for high-volume scenarios.
sudo apt update && sudo apt upgrade -y
sudo apt install -y docker.io docker-compose
docker run -d --name openclaw-automation \
-p 3000:3000 \
-v /opt/openclaw/data:/app/data \
-v /opt/openclaw/workflows:/app/workflows \
--restart always \
openclaw/openclaw:latest
Configure per the official guide.
{
"skills": [
{
"name": "document-processor",
"capabilities": ["ocr", "extraction", "classification", "validation"]
},
{
"name": "approval-engine",
"capabilities": ["routing", "escalation", "sla_tracking"]
},
{
"name": "integration-hub",
"connectors": ["rest_api", "database", "email", "im", "file_system"]
},
{
"name": "ai-decision-engine",
"capabilities": ["classification", "scoring", "recommendation", "anomaly_detection"]
}
]
}
Invoice Processing Automation:
process: invoice_processing
trigger: new_file_in(/inbox/invoices/)
steps:
- extract:
skill: document-processor
action: ocr_and_extract
fields: [vendor, invoice_number, date, line_items, total, currency]
confidence_threshold: 0.9
- validate:
checks:
- vendor_in_approved_list
- amount_within_budget
- no_duplicate_invoice
- tax_calculation_correct
on_failure: route_to_manual_review
- classify:
skill: ai-decision-engine
categories: [capex, opex, recurring, one_time]
assign_cost_center: based_on_vendor_and_category
- approve:
skill: approval-engine
rules:
- amount < 1000: auto_approve
- amount < 5000: manager_approval
- amount < 25000: director_approval
- amount >= 25000: vp_approval
sla: 48_hours
escalation: manager_of_approver
- execute:
on_approved:
- create_payment: erp_system
- update_ledger: accounting_software
- archive_document: cloud_storage
- notify: [requester, finance_team]
Employee Onboarding:
process: employee_onboarding
trigger: new_hire_record_in_hr_system
steps:
- gather_info:
from: hr_system
fields: [name, role, department, start_date, manager, location]
- provision_accounts:
parallel:
- create_email: google_workspace
- create_slack: slack_admin_api
- create_github: github_org_api
- create_jira: atlassian_api
timeout: 30m
- setup_equipment:
skill: ai-decision-engine
determine: equipment_package
based_on: [role, department, seniority]
order_from: it_asset_management
- generate_welcome_kit:
skill: document-processor
create:
- welcome_letter: personalized
- first_week_schedule: based_on_role
- training_plan: based_on_department
- mentor_assignment: based_on_team
- schedule_meetings:
- orientation: day_1_morning
- team_intro: day_1_afternoon
- manager_1on1: day_2
- hr_benefits_review: day_3
- notifications:
- manager: "New team member {{name}} starting {{start_date}}"
- team: "Welcome {{name}} to the team!"
- it: "Equipment order for {{name}}"
- facilities: "Desk setup for {{name}} at {{location}}"
Support Ticket Routing:
process: ticket_routing
trigger: new_support_ticket
steps:
- analyze:
skill: ai-decision-engine
extract: [category, severity, product, customer_tier]
sentiment: detect
- enrich:
lookup_customer: crm_database
get_history: previous_tickets
check_sla: contract_terms
- route:
rules:
- severity critical + customer enterprise:
→ team: senior_engineers
→ sla: 1_hour
→ notify: engineering_manager
- category billing:
→ team: finance_support
→ sla: 4_hours
- category technical + product in [api, sdk]:
→ team: developer_support
→ sla: 8_hours
- default:
→ team: general_support
→ sla: 24_hours
- auto_suggest:
search: knowledge_base
provide: top_3_solutions
to: assigned_agent
monitoring:
dashboards:
- active_processes: real_time
- sla_compliance: hourly
- bottleneck_detection: daily
- cost_per_process: monthly
metrics:
- process_completion_rate
- average_cycle_time
- automation_rate (ai_vs_human)
- error_rate
- cost_savings
alerts:
- sla_breach_imminent → escalate
- error_rate > 5% → investigate
- queue_depth > 100 → scale_warning
| Process | Manual Cost | Automated Cost | Savings |
|---|---|---|---|
| Invoice Processing | $15/invoice | $0.50/invoice | 97% |
| Employee Onboarding | $800/hire | $50/hire | 94% |
| Ticket Routing | $5/ticket | $0.10/ticket | 98% |
| Infrastructure | — | $15/month | — |
Build your intelligent process automation system on Tencent Cloud Lighthouse.