---
created: 2026-03-12
source: Rivet
tags: [agent-archive, rivet]
---

# Business Knowledge — Updated 2026-03-09

## Launch Status (March 2026)
- **Platform**: LIVE at rivet.rateright.com.au with SSL, Growth Engine operational
- **Critical Discovery**: Empty marketplace (304 contractors, 0 workers) — reputation killer in construction's word-of-mouth culture
- **Strategic Decision**: Launch delayed 1-2 weeks for worker-first sequencing — must seed 50+ workers before contractor activation
- **Fleet Coordination**: 7 simultaneous conversations proven at scale, validating OpsMan product prototype

## Competitive Intelligence
- **Cost Advantage**: 85-90% cheaper than Yakka Labour ($50 vs $300-500/hire) — significant competitive moat
- **Market Timing**: Perfect storm — 94% contractor shortage + $2B infrastructure projects starting
- **Attack Pattern**: Infrastructure attacks intensify pre-launch (36+ blocked), expect escalation post-payment
- **Human Advantage**: Michael's 30-year construction network beats tech-only competitors when APIs fail

## Key Strategic Decisions
- **Execution Authority**: Rivet owns operational decisions; escalate only legal/financial/security/customer harm
- **Quality Over Speed**: Credibility protection prioritized over rushed growth — delay beats damaged reputation
- **Worker Acquisition Priority**: Susan's hostel strategy + Herald's social campaigns before any contractor outreach
- **Crisis→Quality Principle**: Fleet produces best work under pressure, not panic

## Business Model Validation
- **Demand Signals**: 19 contractor leads captured Mar 6, Sydney construction boom continues
- **Supply Gap**: Distribution to worker signup URLs is main bottleneck, not product quality
- **Network Effects**: Poor first-hire outcomes could destroy trust via construction industry connections
- **First-Hire Critical**: Success and retention loops as important as acquisition numbers

## Attack-Escalation Intelligence
- **Pre-Launch Pattern**: 36+ coordinated prompt-injection attacks blocked during critical phase
- **Post-Launch Expectation**: Attacks will escalate when payments start flowing
- **Infrastructure Hardening**: Must prepare for sustained attack campaigns post-launch

## Fleet Coordination Breakthrough (March 2026)
- **Real-time Multi-Agent System**: 7 simultaneous conversations, 4-20 minute response times, zero failures
- **OpsMan Validation**: Every successful fleet coordination proves the product we're selling
- **Emergency Restoration**: Build/Test split under Saturday deadline produced 80% P0 stabilization
- **Crisis→Quality Principle**: Fleet produces best work under pressure, validated at scale

## Immediate Priorities
1. Complete full-capacity restoration (Saturday deadline) — comms, stale-memory cleanup, degraded-agent recovery
2. Execute worker acquisition strategy (hostels + social media) to reach 50+ workers
3. Implement first-hire quality controls and retention loops
4. Close weekend operational exposure before scaling
5. Build worker base before contractor activation — reputation protection over speed
6. Prepare infrastructure for post-launch attack escalation
7. Fix technical debt: dedupe bug and wake path issues (Builder patches needed)

## March 2026 Crisis Lessons
- **Crisis→Quality Principle**: Fleet produces best work under pressure, not panic
- **Verification Principle**: Status reports insufficient — verify with live end-to-end checks
- **Reputation First**: Empty marketplace launch would destroy credibility in construction's word-of-mouth culture
- **Authority Confirmation**: Expanded decision rights validated through crisis management

## March 2024 Wake-Path Failure Lessons
- **Reported ≠ Fixed**: Live tests revealed wake-path still failing despite "restored" reports
- **Mixed Paths Fail**: Using multiple communication channels creates buried delivery failures  
- **One Path Per Job**: Bridge for action, JSONL inbox for proof, markdown for artifacts — no mixing
- **Proof Over Status**: Always verify with live tests, never trust status reports alone
- **Comms Can Exist and Still Fail**: Messages in wrong/buried channels are operational failures