Manager Training Visual — Revised

HOUSE Equation
Agent Ecosystem Map

Every workflow in the coaching program, floor by floor — who runs it, what an agent already does, and what's still a human call.

Revised August 26, 2026 Supersedes previous August 26, 2026 build (adds trigger tagging) Status Final Layers 5 — Lead Entry → Coach → Manager → Admin → Founder
Governing principle: coaches think and talk; agents draft and administer. Every workflow below is one of four things — an agent runs it end to end, an agent drafts and a human approves, a human owns it outright, or it's not built yet.
Agent runs the step itself, no draft-review loop Hybrid agent drafts, human approves Human no agent involved Deferred unscoped, dependency-gated, or baseline spec only Schedule runs on a recurring cadence Event fires immediately on a system event
Last updated: Aug 26, 2026
What changed since the last build (Aug 26):
  • Added a schedule-vs-event trigger badge to every Cowork/Claude Code agent card across Layers 1–3, plus two new legend entries.
  • Added Prospect Conversion Agent (L1, event-triggered) and Business Plan Optimizer (L2, schedule-triggered) cards to fill gaps the tagging pass surfaced.
  • Added this status indicator and changed-since summary block.
  • Flagged an open conflict on the punch list — MMR Progress Monitor and Scholarship Cohort Warden were proposed as standalone agents in an earlier brief, but the Aug 20 working session folded both elsewhere. Not added as cards pending a decision.
L0

Lead Entry

Two doors in, three pipes feeding them.

Entry points

HOUSE Equation App direct signup, and the free Trailblazers Community.

Lead sources

Coach campaigns (70–80% of volume) · Company campaigns (centrally generated) · Manager campaigns (self-coach or assign out).

Execution

GHL Native Managed Agent executes campaigns; coach or manager approves.

Hybrid
L1

Coach Workflows

Owns the student relationship — first contact through the Acquisition Checklist.

01

Lead Generation & Nurture

GHL managed agent executes, coach approves.

Hybrid
02

Appointment Setting

Appointment Setter Agent — built and live, running ~3–4 days. No student volume through it yet; still in launch/startup phase.

Function 4 (added Aug 24) — Inbound Social/Community Engagement Response. On Facebook: the agent monitors comments and DMs directly, answers questions, and drives toward booking, under the same guardrails as the Social Content Agent (no guaranteed outcomes/timelines, no legal/tax/lending advice). Inside the GHL Community: the agent does NOT monitor directly, to avoid collision with the Community Engagement Agent, which already owns that space. It only activates on a handoff. Handoff mechanic: when the Community Engagement Agent spots a booking-worthy inquiry, it posts a natural, varied acknowledgment-plus-handoff message (not a fixed script) before the Appointment Setter Agent takes over that thread — avoids a jarring silent context-switch for the prospect.

AgentFunction 4 — social/community inbound
03

First Strategy Session

Fully human. Relationship-building plus the AD diagnostic.

Human
04

Weekly Check-Ins

Weekly Check-in Triage Agent drafts response and rating, coach approves.

HybridSchedule — weekly
05

Ad Hoc Student Messages

Agent drafts and triages for escalation, coach approves.

Hybrid
06

Cash Flow Audit Notes

Cash-Flow Audit Copilot drafts notes in a casual tone, coach edits for tact.

HybridEvent — fires on statement submission
07

Ecosystem Activation Matching

Coach decides role fit — a personality read stays human judgment. Agent drafts the explainer after.

Human decides Hybrid drafts
08

Checklist Release Pacing

Fully human judgment call.

Human

Social Content Agent

Built and live since August 22, 2026 — operates inside GoHighLevel, drafting daily Facebook/GHL community posts. Not tied to one numbered workflow above; supports Layer 0/1 lead nurture broadly.

Agent

Email Campaign Agent

Identified alongside the Social Content Agent, not yet scoped as its own build. Job: consistency and comprehension engine across the full ecosystem — not just lead nurture. Seven audience tracks identified (Aug 24 session): cold leads; students/community by Track 1 vs Track 2; students/community by passive role (tenant-buyer partner, landlord, silent partner) vs active role (Real Estate Collaborator, Deal Scout, Disposition Partner, Acquisition Partner) — passive gets core-message reinforcement, active gets structured deal education; coaches; managers; strategic partners split by RE/mortgage professionals vs vendors (mirrors the existing Coach Prospecting Strategy pitch-logic split). Unifying root narrative across every track: everyone is either directly acquiring assets and building passive cash flow, or supporting someone else's journey to do the same — same destination (Tier 1/2/3 financial freedom, active-to-passive cash flow transition via real estate), same standardized mental model (Leap Frog Method). Sequence structure and cadence still to be designed.

Deferred

SMS Nurture Agent

Identified alongside the Social Content Agent and Email Campaign Agent as a separate, distinct build (different discipline, different risk profile — a bad SMS blast is more damaging than a mediocre post). Not yet scoped.

Deferred

Prospect Conversion Agent

Watches new registrations, fixes the prospect→student attribution and coach-assignment gap (a known fragile flow), auto-assigns to the right coach, triggers the conversion workflow, and notifies the coach. Not yet scoped as its own build.

DeferredEvent — fires on new registration
L2

Manager Workflows

Owns the coach network — onboarding, lead assignment, escalation, certification sign-off.

01

Self-Coaching

Identical to Layer 1 — run directly with the manager's own students.

= Layer 1
02

Weekly Coach Check-Ins

Anchored on the coach's Business Plan pace vs. goal.

Human today

Activity Tracker is now built (Base44). Pace Flagging Agent is unblocked — ready to scope whenever prioritized, not yet built.

Unblocked, not builtSchedule — weekly
03

Campaign Management

Self-coach or assign to network; runs through the GHL managed agent — no manager-specific gap.

Hybrid
04

Escalation Handling

Two distinct paths, not one inbox — see diagram below.

HybridEvent — fires on coach flagging an escalation
05

Community Rotation

Shared calendar and group chat. No agent needed.

Human
06

Community Engagement Agent

Renamed from "Community Moderation." Primary job is engagement — welcoming new members, orienting them, guiding them to the right channels, nudging participation. Moderation (bad behavior, disputes, spam) is a secondary function folded into the same agent, not a separate build.

Scoped, not built
07

Cash Flow Certification Decision

Not the manager. Admin/founder only — the manager's role stops at reading the coach's notes upstream of that decision.

Admin/Founder only
08

Coach Onboarding Nudges

Folded into Weekly Coach Check-ins (item 02) for now — becomes a watch-condition on that same agent later, not a new build.

Folded inSchedule — weekly

Business Plan Optimizer

Analyzes a coach's funnel inputs on the existing Business Plan calc engine and suggests conversion-rate or pricing changes to maximize annual income — a "what-if" copilot layered on the current calculator. Not yet scoped.

DeferredSchedule — monthly
Coach hits an out-of-lane situation Escalation Routing Agent first-pass triage + draft route Path A — Complexity tax, legal, out-of-lane Path B — Expertise live deal-level question Manager or Admin ± outside professional / directory Real Estate Collaborator via RPM Partnership Directory Human approves before anything routes
Escalation Handling (L2·04) — same triage agent, two different destinations. Expertise questions go to a Real Estate Collaborator, not up the management chain.
L3

Admin Workflows

Owns platform integrity — certification gates, audits, and everything that feeds founder-level oversight.

Gate 1 — Coach Intake values + capacity survey Coach Intake Assessment Agent clean pass borderline / fail Gate 2 — CHEC Exam 80% pass threshold Admin + vouching manager joint decision, not auto-reject CHEC Auto-Grant Agent fully mechanical cert + badge + notify manager
Coaches Tab (L3·01) — two gates, one mechanical, one human-joined. A clean intake score never touches a person until certification is already earned.
Event — Coach Intake Assessment Agent fires on new coach application Event — CHEC Auto-Grant Agent fires on passing exam score
02

Managers Tab

Oversight only — network health.

Human
feeds Founder Oversight Agent
03

Students Tab

Oversight only — platform-wide roster health.

Human
feeds Founder Oversight Agent
04

MMR Sessions

Trend and KPI watch.

Human
feeds Founder Oversight Agent
05

Unassigned Prospects

Prospect Assignment Recommendation Agent proposes coach, manager, or self-coach via round robin + capacity; admin or manager approves.

HybridEvent — fires on admin inviting a student
06

Scholarship Recipes & Queue

No agent — low-frequency configuration.

Human
07

Cash Flow Audits

Cash-Flow Audit Copilot (extended) drafts summary, report, and a recommended decision; admin/founder makes the call. Matched Real Estate Collaborator gets the outcome on all three results — no vote.

HybridEvent — fires on statement submission
08

Remedy Library

Remedy Intelligence Agent (extended): live-session recommendations, plus flags high-quality coach remedies as promotion candidates for admin review.

Trigger: in-session, inline — not a scheduled batch or event webhook, so it doesn't take a schedule/event badge.

Hybrid
09

Session Outline

Session Outline Optimization Agent (new) mines call transcripts against conversion outcomes, proposes script revisions. Full admin approval on every change.

HybridSchedule — weekly/biweekly batch
10

Onboarding

Execution is the manager's job; admin's view is passive.

Human
feeds Founder Oversight Agent
11

Escalations

Escalation Routing Agent (extended with a type classifier): hierarchy default, expertise path, and a vendor path — now unblocked pending minimum bench size. The HOUSE Equation Partner Program (see Partner Program Overview + Partner Leave-Behind docs) has been initiated to populate the RPM Partnership Directory with tax, title, and insurance vendor listings. Routing logic is already defined: situational/expertise-specific, e.g. a tax question routes to a tax professional, an insurance question to an insurance agent — giving vendors relationship-building and visibility in the community as the incentive to participate.

Hybrid Vendor path unblocked Event — fires on coach flagging an escalation
12

Feedback

Passive student-on-coach signal.

Human
feeds Founder Oversight Agent
13

Change Log / Announcements / Guide

No agent — authoring tools.

Human
L4

Founder-Level Oversight

Baseline spec, designed August 21, 2026 — build deferred until the ecosystem hits the threshold below, not until any data-volume mark. Same data pipeline, two different jobs, deliberately kept separate.

Build threshold (decided August 23, 2026): the full ecosystem lands at roughly 10–12 distinct agents. Founder Oversight Agent specs get built out once about half of them (5–6) are live — gated on the builds themselves being done, not on data volume, since the builds are what determine what data exists and in what shape.

Live — 2 of ~10–12

Appointment Setter Agent · Social Content Agent.

Agent

Not yet built — ~8–10

Weekly Check-in Triage · Ad Hoc Student Message · Cash-Flow Audit Copilot · Pace Flagging (unblocked) · Escalation Routing · Community Engagement · Coach Intake Assessment · CHEC Auto-Grant · Prospect Assignment Recommendation · Session Outline Optimization · Remedy Intelligence · Email Campaign Agent · SMS Nurture Agent · Prospect Conversion Agent · Business Plan Optimizer. Some are extended, dual-layer versions of the same build.

Deferred

Agent A

Founder Oversight Agent

Job
Monitor every other agent's performance and macro KPIs, detect anomalies and workflow gaps.
Cadence
Hybrid — weekly scheduled digest, plus real-time alerts as anomalies are detected.
Output
Status report + drafted recommendations the founder can act on.
Data in
MMR sessions, Cash Flow Coach friction points, student activity patterns, coach/manager Activity Tracker pace data (Activity Tracker now built), plus performance output from every Layer 1–3 agent.
Named watch-item (Aug 24)
Function 4 handoff quality between the Community Engagement Agent and the Appointment Setter Agent — conversion rate, over/under-flagging of "booking-worthy" inquiries, and handoff smoothness. Explicitly named rather than left implied under general agent monitoring. Until the Founder Oversight Agent is built, Eric is the interim human check via spot-checking conversation logs.
Reports to
Founder only — never surfaces directly to managers.
Open item
Anomaly thresholds not yet defined — deferred until real data accumulates to calibrate against.
Deferred — baseline spec

Agent B

Opportunity / Innovation Agent

Job
Same data access as Agent A, scanning for patterns that suggest new tools, features, or efficiencies for students, coaches, and managers.
Output
Build proposals for founder review before anything gets built — not status reports.
Why separate
Status reporting and product ideation call for different review muscles; one agent shouldn't carry both.
Deferred — baseline spec

Standing Open Items

The punch list — outstanding across every layer above, not tied to one.

  1. Resolved · Aug 23  Coach/student strategy-comfort safeguard. Original framing — a coach incentivized to accelerate a student's Acquisition Date against their interest — was reconsidered as impractical, since roadblocks are time- and situation-based and can't be arbitrarily compressed for personal gain. The real risk instead: a coach steering a student toward a newer, objectively faster strategy the student isn't yet comfortable with, when they'd naturally gravitate toward a slower, more familiar one. Safeguard: coaching calls already run on recorded GHL phone lines — an agent can monitor transcripts for signs of student discomfort with a recommended strategy. No new build required; rides on existing call-recording infrastructure.
  2. Community Engagement Agent (renamed from Community Moderation). Scope is now defined — see L2·06: engagement/onboarding is the primary job, moderation is secondary. Not yet built.
  3. Activity Tracker. Built — completed in Base44. The Pace Flagging Agent (L2·02) is now unblocked; not yet scoped or built.
  4. Founder Oversight Agent, both halves. Baseline spec is done. Build threshold is now defined as roughly half of the total ecosystem agents (5–6 of 10–12) live — not a data-volume threshold. See Layer 4.
  5. Resolved · Aug 24  Vendor escalation path (L3·11). RESOLVED/UNBLOCKED as of Aug 24 — HOUSE Equation Partner Program now initiated to populate the directory with vendor listings. Remaining condition: minimum bench size before the routing agent has a meaningful destination pool.
  6. Blocked · pending Base44  GHL segmentation/tagging structure. The GHL tag structure for Track (Track 1 owner-occupant / Track 2 rental-investor), lifecycle stage (New Lead → Contacted → Link Sent → Account Created → Appointment Booked → Session Held → Checklist Released), and coach/manager ownership is not yet buildable — it depends on how Base44 actually stores these three fields today. Prompt sent to Base44 AI (Aug 26) requesting exact field names, entities, and null-ability for track assignment, lifecycle stage, and coach/manager ownership — see "Base44 AI Prompt - GHL Tagging Alignment." Dependency, not a build gap: holds until that answer comes back.
  7. Schedule vs. event trigger architecture (Aug 26). Cowork/Claude Code agents split into two trigger types: event-based (fires on a system event via an intermediary like Zapier/n8n calling a Claude Code Routine's API trigger) and schedule-based (runs on a fixed cadence via Cowork Scheduled Tasks or Claude Code Routines). GHL-native agents (Social Content, Email Campaign, SMS Nurture, Appointment Setter, Community Engagement) are out of scope for this framing — they use GHL's own trigger system.
  8. Open · needs a decision  MMR Progress Monitor / Scholarship Cohort Warden — naming conflict (Aug 26). Both appear as standalone Tier 2 agent proposals in the Aug 18 End-to-End App Brief. The Aug 20 Admin Workflow session decided against splitting them out: MMR trend/KPI watching folds into the Founder Oversight Agent's "Data in" feed (see Layer 4), and Scholarship Recipes/Queue is "no agent needed — low-frequency configuration" (L3·06). Not added as standalone cards on this map — flagging the conflict rather than picking a side.
HOUSE Equation — Full Workflow & Agent Ecosystem Map · Manager Training Visual · Revised August 26, 2026 (supersedes previous August 26, 2026 build)