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

# RateRight Agent Connection Map

*Systems architecture mapping how all RateRight agents connect — what they produce, consume, and depend on.*
*Created: 2026-02-07*

## Agent Profiles

### 1. Builder Agent
**What it PRODUCES:**
- Code changes to the-50-dollar-app (Next.js + Supabase)
- Bug fixes and feature implementations
- Database migrations and schema updates
- API endpoints and UI components
- Security audit fixes
- Deployment-ready builds

**What it CONSUMES:**
- Product specifications from Researcher
- UI/UX research and design guidelines
- Security audit reports from SysOps
- TODO.md task assignments from Rivet
- ABR API documentation for verification features

**What agents does it DEPEND ON?**
- Researcher (for specs and market research)
- SysOps (for deployment and infrastructure)
- Rivet (for task coordination)

**What agents DEPEND ON IT?**
- Sales (needs working app to demo to leads)
- Content (needs features to write about)
- SiteOps (uses app for daily logs)

---

### 2. Sales Agent
**What it PRODUCES:**
- Lead conversions and customer acquisitions
- Call logs and conversation notes
- SMS sequences and outreach campaigns
- Lead intelligence and buying signals
- Revenue ($50 per hire)
- Market feedback from prospects

**What it CONSUMES:**
- Competitor research from Researcher
- Lead lists from Growth Engine API
- SMS scripts and sequences from Content
- AI call prep briefs from Growth Engine
- Product features updates from Builder

**What agents does it DEPEND ON?**
- Researcher (competitor analysis, market intel)
- Content (SMS scripts, brand voice)
- Builder (working product to sell)
- SysOps (Growth Engine API availability)

**What agents DEPEND ON IT?**
- Researcher (needs feedback on what's working)
- Content (needs real objections to address)
- Rivet (needs revenue data for planning)

---

### 3. Researcher Agent
**What it PRODUCES:**
- Competitor analysis reports
- Market sizing and industry data
- Regulatory compliance research
- Product specifications and feature requirements
- UX/UI research and best practices
- API integration documentation

**What it CONSUMES:**
- Sales feedback on market response
- Builder questions about technical feasibility
- Legal requirements from compliance needs
- Industry trends and government data

**What agents does it DEPEND ON?**
- Sales (for real-world market feedback)
- SysOps (for access to web research tools)

**What agents DEPEND ON IT?**
- Builder (needs specs before building)
- Sales (needs objection handling data)
- Content (needs topics to write about)
- Legal (needs regulatory research)

---

### 4. SysOps Agent
**What it PRODUCES:**
- System stability and uptime
- Security configurations and fixes
- API key management and rotation
- Infrastructure monitoring and alerts
- Configuration backups and recovery
- Gateway maintenance and updates

**What it CONSUMES:**
- Error logs from all systems
- Security audit findings
- Configuration requirements from other agents
- Infrastructure needs from scaling demands

**What agents does it DEPEND ON?**
- All agents (for error reports and issues)

**What agents DEPEND ON IT?**
- All agents (need stable infrastructure)
- Builder (needs deployment environment)
- Sales (needs Growth Engine API)
- Researcher (needs web search tools)

---

### 5. SiteOps Agent
**What it PRODUCES:**
- Daily work logs and timesheets
- Site progress documentation
- Issue tracking and delay reports
- Photo documentation of work
- Project coordination notes
- Concrete pour schedules

**What it CONSUMES:**
- Site instructions from foremen
- Engineer specifications
- Weather and delay information
- Material delivery schedules

**What agents does it DEPEND ON?**
- Builder (for Notion integration features)
- SysOps (for system stability)

**What agents DEPEND ON IT?**
- Rivet (needs daily updates for morning brief)
- Content (can use site stories for content)

---

### 6. Rivet (Main Agent / COO)
**What it PRODUCES:**
- Task coordination and prioritization
- Strategic decisions and planning
- Daily morning briefs
- TODO.md maintenance
- Agent spawning and orchestration
- System-wide coordination

**What it CONSUMES:**
- Updates from all agents
- Michael's evening calls and priorities
- System health metrics
- Revenue and pipeline data
- Market conditions and feedback

**What agents does it DEPEND ON?**
- All agents (for status updates and completion reports)

**What agents DEPEND ON IT?**
- All agents (need task assignments and coordination)

---

## Connection Flow Map

```
Research → competitor data → Sales (for objection handling)
Research → market data → Content (for blog topics)
Research → UX findings → Builder (for implementation)
Research → regulatory info → Legal (for compliance)

Sales → lead feedback → Research (what's working, what's not)
Sales → revenue data → Rivet (for planning)
Sales → objections → Content (for script improvements)

Builder → new features → Content (what to market)
Builder → new features → Sales (what to sell)
Builder → product updates → All agents (status updates)

Content → SMS scripts → Sales (for campaigns)
Content → blog posts → Research (for SEO/data)
Content → brand voice → All agents (for consistency)

SysOps → system health → All agents (infrastructure status)
SysOps → security fixes → Builder (for implementation)

SiteOps → daily logs → Rivet (for morning brief)
SiteOps → site issues → Content (for authentic stories)

Michael (evening call) → priorities → Rivet → TODO.md → All agents
```

## Identified Issues

### Broken Connections
1. **Research → Sales**: Research produces competitor pricing data, but Sales doesn't have automated access to update objection handling scripts
2. **Sales → Builder**: Sales collects feature requests from leads, but no systematic way to feed into product roadmap
3. **SiteOps → Builder**: SiteOps documents real construction issues, but Builder doesn't get user stories for feature development

### Missing Connections
1. **Content → Sales**: Content creates marketing materials that Sales could use in conversations
2. **Legal → Sales**: Legal requirements should inform Sales scripts (compliance messaging)
3. **SiteOps → Research**: Real site experience could validate research assumptions about user needs

### Bottlenecks
1. **Rivet**: Everything flows through the main agent - single point of failure for coordination
2. **Builder**: All development work flows through one agent - could slow feature delivery
3. **Michael**: Evening calls create daily bottleneck for priority setting

### Redundancies
1. **Multiple agents doing research**: Researcher does market research, but Sales also researches leads, Content researches topics
2. **Status updates**: Each agent reports to Rivet separately, could be consolidated

### Orphaned Outputs
1. **SiteOps photos**: Daily site photos aren't used for marketing or product development
2. **Sales call recordings**: Rich conversation data isn't analyzed for patterns
3. **Research reports**: Some research sits unused if no agent immediately needs it

### Starved Inputs
1. **User validation**: No agent is getting real user feedback from the app
2. **Competitive intelligence**: No systematic tracking of competitor moves
3. **Market trends**: No ongoing monitoring of construction industry changes

## Recommendations

### New Connections to Add
1. **Automated Research → Sales feed**: Create API endpoint for Sales to pull latest competitor intel
2. **Sales feedback loop**: Formal process for Sales to submit feature requests to Builder
3. **Site story pipeline**: SiteOps site issues automatically become user stories for Builder
4. **Content library**: Central repository where Content saves materials for Sales use
5. **Legal compliance alerts**: Automatic notifications to Sales when compliance requirements change

### Information That Should Flow Automatically
1. **Build status → Sales**: When Builder completes features, Sales gets automatic notification
2. **Lead feedback → Research**: Sales tags objections, automatically aggregated for Research
3. **System health → All**: SysOps issues broadcast to all agents
4. **Daily summaries**: Automated digest of each agent's work sent to all

### How to Make Agents Aware of Outputs
1. **Shared memory system**: All agents check `memory/plans/` before starting work
2. **Notification system**: Automated alerts when relevant research is completed
3. **Cross-agent tags**: Tasks tagged with multiple agent codes when relevant
4. **Weekly sync**: Automated summary of cross-agent dependencies and outputs

### Implementation Priority
1. **High**: Fix broken connections (Research→Sales, Sales→Builder)
2. **Medium**: Add missing connections (Content→Sales, Legal→Sales)
3. **Low**: Address redundancies and optimize flows

---

*This map should be reviewed weekly and updated as agents evolve and new connections are identified.*