Vapora/examples/real-world/README.md

Real-World Examples

Production scenarios demonstrating VAPORA solving business problems.

Use Cases

01: Code Review Pipeline

File: 01-code-review-workflow.rs

Automated multi-agent code review with cost optimization.

Business Value:

• Review 50 PRs/day consistently
• 99% accuracy matching human review
• $500+/month savings vs manual review
• Instant feedback to developers

Architecture:

1. Ollama (free) - Static analysis
2. GPT-4 ($10/1M) - Quality review
3. Claude ($15/1M) - Architecture review

Run:

cargo run --example 01-code-review-workflow

02: Documentation Generation

File: 02-documentation-generation.rs

Automatically generate and maintain API documentation.

Business Value:

• Docs always in sync with code
• 2-3 week lag → instant updates
• 99%+ accuracy vs manual docs
• $1000+/month savings

Architecture:

1. Ollama - Code analysis
2. Claude - Doc writing
3. GPT-4 - Quality check

Run:

cargo run --example 02-documentation-generation

03: Issue Triage

File: 03-issue-triage.rs

AI-powered issue classification and routing.

Business Value:

• 200+ issues/month triaged instantly
• Consistent classification criteria
• 95% accuracy
• $1000+/month savings (vs 20 hours manual work)

Architecture:

1. Ollama (free) - Initial classification
2. Claude ($0.08) - Detailed analysis when needed

Run:

cargo run --example 03-issue-triage

Cost Analysis Pattern

Real-world examples use cost-aware multi-stage pipelines:

Stage 1 (Cheap): Ollama/GPT-4 for routing
  ↓
Stage 2 (Premium): Claude only when needed
  ↓
Stage 3 (Validation): GPT-4 for quality check

This achieves:

• 95%+ accuracy at 10% of premium cost
• Fast iteration with feedback
• Controlled budget

Key Metrics

Each example demonstrates:

Metric	Typical Value
Accuracy	95%+ vs manual
Speed	10-100x faster
Cost	1-10% of manual
Coverage	100% automated

Adapting for Your Use Case

1. Choose pipeline stages: Cheap → Medium → Premium
2. Adjust triggers: When to escalate to next stage
3. Set budgets: Per-role monthly limits
4. Monitor quality: Track accuracy vs manual baseline

Next Steps

• Implement case-specific routing rules
• Integrate with your systems (GitHub, Jira, etc.)
• Monitor and adjust thresholds
• Expand to additional use cases

---

Navigate to ../ for more examples.