Meet your Village agents.
Each task spawns a focused specialist.
When your team makes a request, a fresh agent spawns. It loads the master context for The Village plus the current context relevant to the task, executes the request, reports back, and ends. The next request spawns the next agent. Your team directs. The agents execute. You stay in charge of every task.
The Briefing Agent
Works the ownership briefing surface. Spawned when your team requests a project read.
Tasks it executes on request
- Pulls a structured read across project management, accounting, leasing, and field reporting systems
- Drafts a one-page summary of where Phase 1 stands across budget, schedule, and pre-leasing positioning
- Compares last week against this week, flags movement, drafts a brief
- Surfaces decisions awaiting principal review with context attached
- Drafts a partner-facing memo on the quarter's project position when requested
How your team directs it
- Triggered by name from the Ownership Briefing surface, by menu selection, or by structured form input. Never by free-form prompt.
What it does not do
- Does not approve decisions, dispatch work, release capital, or modify project scope. Every output is a draft for principal review.
How it ends
- Reports back to the principal with a structured outcome record. The agent ends. The next briefing request spawns a fresh instance.
And the others.
Same pattern, different specialties. Each one spawns on request, executes one task, reports back. Listed in lifecycle order from buildout through operations.
The Capital Agent
Works the capital and draws phase.
Example task: Pulls last month's draws against the underwritten budget, flags variance over a defined threshold, drafts a structured read.
The Schedule Agent
Works the project schedule phase.
Example task: Reads the project baseline against the current schedule, drafts a slip report on critical path items with downstream impact.
The Trades Agent
Works the trades and vendors phase.
Example task: Pulls inspection status across active trades, flags items over five days pending, drafts a status note.
The Decision Agent
Works the ownership decision phase.
Example task: Assembles a decision package for principal review with supporting context, source documents, and recommended action attached.
The Leasing Agent
Works the pre-leasing and tenant cultivation phase.
Example task: Drafts a retail prospect status read by category, flags movement since last review, surfaces decisions for the leasing team.
The Operating Agent
Works the operating tenancy phase. Activates as Phase 1 buildings approach delivery.
Example task: Spawns the make-ready, notice, listing, deal, and onboarding agents that handle the operating lifecycle once buildings are live.
The Operating Agent activates as Phase 1 approaches delivery, projected Q2 2028.
Why this works.
Each agent fires with full context loaded. That is why the output is reliable. Built once during the engagement, revised as the project evolves.
Master context layer
Loaded into every agent on spawn. Persistent across the operation.
- Project structure, phase plan, and capital stack as a structured knowledge base, not free text
- Organizational hierarchy with role-based access boundaries across principal, partner, asset manager, leasing, and finance
- Terminology, asset codes, and naming conventions specific to the family operation
- Document sources mapped by authority, defining which document is canonical for which question
- Versioned and revisable, with a complete audit trail on every change
Orchestration layer
Defines how every agent reads, reasons, and routes.
- Source-of-truth hierarchy across project management, accounting, leasing, and field tools, defined per question type
- Conflict resolution rules when two systems disagree, configured during onboarding
- Confidence thresholds that determine when an agent answers directly versus surfaces a question to the principal
- Recurring briefing patterns such as weekly ownership reads, monthly capital position, and quarterly partner updates, defined as standing prompts
- Escalation routing for items above defined thresholds, including variance flags, schedule slips, and capital requests
Execution layer
Defines the scoped tasks each agent can execute.
- Pre-built scripts for recurring scoped actions, including notification routing, document generation, data export, and status update propagation
- Each script has defined inputs, defined outputs, defined permissions, and a defined audit record
- Triggered by name, by menu selection, or by structured form input, never by free-form prompt
- Idempotent and reversible where possible, with destructive actions requiring explicit principal confirmation
- Reports back to the principal on completion with a structured outcome record
Integration layer
Defines what every agent reads from. Nothing replaced.
- Read-only API connections to project management, accounting, leasing, and document systems
- Structured intake from email, calendar, document folders, and exports where APIs are not available
- All data stays in your systems of record. The agents hold no canonical data
- Connection health monitored continuously, with source failures surfaced as confidence flags on affected answers
- Onboarding integration scoped per system during the initial engagement
Each request spawns a specialist. Each specialist runs one task. Your team directs. The system reports back. Trust compounds as your team learns where each agent fires reliably, and the agents get sharper as the master context grows. You stay in charge of every task.