Who uses the system and what AI participates.
SpineWorkspace serves every role in the organization. Each role gets a domain-specific projection of the workbench. AI agents work alongside humans, never replacing them.
Design Principles
- Humans are core actors — AI augments, never replaces
- One Twin per user — Each person has their own AI partner
- Role-scoped — Everything is scoped to the user's role
- Coworker is ambient — Always present, never intrusive
- Domain agents are specialized — Deep expertise in one area
Human Roles
Executive Leader
| Aspect | Description |
|---|
| Focus | Strategic KPIs, organizational health, high-stakes decisions |
| Workbench | Executive dashboard, decision center, reports |
| Twin interaction | Strategic analysis, risk assessment, recommendation |
| Key metrics | Revenue, pipeline, customer health, team performance |
Sales Manager
| Aspect | Description |
|---|
| Focus | Pipeline, opportunities, forecasts, team performance |
| Workbench | Sales pipeline, opportunity views, forecast reports |
| Twin interaction | Deal analysis, next-best-action, pipeline review |
| Key metrics | Pipeline value, win rate, average deal size, cycle time |
Account Manager
| Aspect | Description |
|---|
| Focus | Customer health, renewals, expansion, relationships |
| Workbench | Account views, health scores, renewal tracking |
| Twin interaction | Health analysis, risk detection, expansion opportunity |
| Key metrics | Health score, NPS, renewal rate, expansion revenue |
Finance Analyst
| Aspect | Description |
|---|
| Focus | Revenue, billing, collections, financial reporting |
| Workbench | Financial dashboards, billing views, collection queues |
| Twin interaction | Revenue analysis, billing anomaly, collection priority |
| Key metrics | Revenue recognition, DSO, collection rate, billing accuracy |
Support Lead
| Aspect | Description |
|---|
| Focus | Tickets, SLAs, escalations, customer satisfaction |
| Workbench | Ticket queue, SLA dashboard, escalation tracker |
| Twin interaction | Ticket triage, SLA prediction, escalation routing |
| Key metrics | SLA compliance, resolution time, CSAT, ticket volume |
Any Role / Customer
| Aspect | Description |
|---|
| Focus | Role-specific work, domain-specific views |
| Workbench | Customizable work surface, domain projections |
| Twin interaction | Contextual assistance, work support |
| Key metrics | Role-specific KPIs |
AI Agent Taxonomy
AI Twin (Per User)
| Aspect | Description |
|---|
| Scope | Per-user, role-scoped |
| Behavior | Reasons, plans, suggests — never acts without approval |
| Activation | Explicit (Four Buttons) or pipeline-triggered |
| Memory | Persistent memory scoped to user's domain |
| Interaction | Four Buttons: Store, Ask, Assign, Approve |
Twin Capabilities
| Capability | Description |
|---|
| Context assembly | Builds understanding from the Spine |
| Reasoning | Analyzes situations and options |
| Planning | Proposes step-by-step plans |
| Drafting | Creates content and communications |
| Research | Investigates questions and patterns |
| Prediction | Forecasts outcomes and trends |
Twin Rules
- Twin never acts without explicit approval
- Twin prefers silence over low-value commentary
- Twin is scoped to the user's role domain
- Twin memory is governed by the same policies as the Spine
AI Coworker (Context Companion)
| Aspect | Description |
|---|
| Scope | Ambient, context-wide |
| Behavior | Observes, gathers, supports — never proposes |
| Activation | Continuous observation |
| Memory | Ephemeral, session-scoped |
| Interaction | Contextual surfacing, communication support |
Coworker Capabilities
| Capability | Description |
|---|
| Context observation | Watches active work without interfering |
| Information retrieval | Pulls relevant data on demand |
| Communication support | Helps messages get to the right people |
| Context alignment | Ensures everyone has the same understanding |
Coworker vs. Twin
| Aspect | Twin | Coworker |
|---|
| Scope | Full reasoning and execution | Context gathering and support |
| Activation | Explicit or pipeline-triggered | Continuous observation |
| Autonomy | Proposes, waits for approval | Supports, doesn't propose |
| Depth | Deep analysis and planning | Lightweight context surfacing |
| Memory | Persistent, role-scoped | Ephemeral, session-scoped |
Domain Agents
| Agent | Domain | Capabilities |
|---|
| Sales Agent | Sales | Pipeline analysis, deal scoring, next-best-action |
| Finance Agent | Finance | Revenue analysis, billing review, collection priority |
| Support Agent | Support | Ticket triage, SLA prediction, escalation routing |
| Account Agent | Account Success | Health analysis, renewal risk, expansion opportunity |
| Operations Agent | Operations | Workflow optimization, resource allocation |
Domain Agent Rules
- Domain agents are scoped to their business domain
- Domain agents can reason and recommend within their domain
- Domain agents escalate cross-domain work to the orchestrator
- Domain agents feed learnings back to the Spine
Specialist Agents
| Agent | Specialty | When Used |
|---|
| Research Agent | Deep investigation | When context is insufficient |
| Analysis Agent | Quantitative analysis | When numbers need interpretation |
| Drafting Agent | Content creation | When output needs to be produced |
| Integration Agent | System integration | When external systems need interaction |
Specialist Agent Rules
- Specialist agents are invoked on demand
- Specialist agents receive focused context for their task
- Specialist agents return results to the calling agent
- Specialist agents do not maintain persistent state
System Agents
| Agent | Function |
|---|
| Orchestration Agent | Coordinates multi-step workflows |
| Scheduling Agent | Manages time-based triggers |
| Monitoring Agent | Observes system health |
| Reconciliation Agent | Verifies execution outcomes |
| Memory Agent | Manages organizational memory |
System Agent Rules
- System agents operate at the platform level
- System agents are not directly visible to users
- System agents enforce governance and policies
- System agents are fully auditable
Agent Interaction Model
Human → Twin → Domain Agent → Specialist Agent → System Agent → Execution → Reconciliation → Memory
Interaction Rules
- Humans initiate or approve all actions
- Twins reason and propose, humans decide
- Domain agents specialize within their scope
- Specialist agents are invoked on demand
- System agents coordinate and govern
- Every interaction is traced in the Spine