INTERNAL · JASON / DRE / ED · JULY 2026

ATLAS

AI Studio × Atrium — the map, not the weeds.

One visual summary of what we're building, how the big platform and the HOCATT consolidation fit together, and how to review the 13 PRDs without drowning in them.

13
module PRDs
18 → 1
HOCATT systems
2
tracks, one bridge
Jul 31
Atrium wrapper live
4–5 mo
AI Studio to replicable
60/40
the scope philosophy
01 · What we're building

Three layers, one platform

Ed's key architecture call: every client gets their own fully isolated instance — own database, own infrastructure. Nothing is shared between clients. "Multi-tenant" exists only at the top, where Result Labs runs the machinery that stamps instances out.

1 · Control Plane someday

Result Labs' master panel. A customer signs up, pays, and the system auto-provisions their isolated Command Center — then watches, upgrades, and bills the whole fleet.

WHO SEES IT: Result Labs only

2 · Command Center the deployable base

The per-client product: one trusted view of sites, analytics, CRM, ads, content, bots and AI agents — finds the leaks, proves what works, turns findings into profitable action. The 13 PRDs describe this layer.

WHO SEES IT: each client's team, roles-scoped

3 · Systems Merger now

The hard, client-specific work of collapsing existing tools into one portal. Atrium (admin.hocatt.com) is the first one — 18 HOCATT systems become one role-based front door.

WHO SEES IT: HOCATT's team — Lisa logs in and sees only sales
02 · The system at a glance

13 modules, three review tiers

Every PRD in one card. Chips show build triage (now / soon / someday) and size (S–XL as spec'd). Tiers are the recommended review order — spend your energy where feedback changes what gets built first.

Tier 1 — Decide now

These shape both Atrium and AI Studio v1. Review deeply.
PRD-08Command Center

One dashboard mapping every platform a client runs — flags what's broken or wasting money, findings become projects. The productized HOCATT control center.

nowXL
PRD-07CRM Integration

Plug-in CRM connections — GoHighLevel first, read-only — into a clean warehouse with automatic data-quality flags (dupes, fake “$0 won” deals).

nowXL
PRD-02Users & Roles

Logins, roles, invites, and audit trail inside each instance. Must reconcile with Atrium's “few roles, many segments” model — our biggest structural feedback.

nowL
PRD-09AI Agents

Chat + voice agents trained on the client's own content, placeable per role (a sales bot only Lisa sees), with every AI call metered for billing. Faith's future home.

nowsoonXL

Tier 2 — The differentiators

Review at 60/40 depth — what's the lean version?
PRD-05Crawler + MCP

A rebuilt crawler that actually reads modern sites, served to AI assistants through each client's private MCP endpoint. Ask: build vs buy?

soonXL
PRD-06Analytics + Bot Tracking

Privacy-friendly analytics that also shows which AI crawlers are reading your site — “ChatGPT read your pricing page.” Really two products in one PRD.

soonXL
PRD-04Content Hub

A per-client content brain: every page and doc as clean, versioned, searchable markdown that agents can cite. The knowledge base under everything AI.

soonL
PRD-03Sites + SEO Suite

Register and verify each site, install tracking, connect Google — plus the full ported SEO toolkit. The registry is now; full SEO parity is negotiable.

nowsoonXL

Tier 3 — Machinery & post-v1

Skim — except Dre reads PRD-01 closely. Watch the two “ships with v1” flags.
PRD-01Provisioning + Fleet

The one-button machinery that stamps out, upgrades, bills, and retires client instances. Ed's own roadmap says: hand-provision the first clients, build this later.

somedayXL
PRD-12Dashboards & KPIs

Client-built dashboards, calculated KPIs, scheduled branded reports — the Databox answer. Its “metric registry” foundation ships with v1.

Phase A = v1rest laterXL
PRD-11Integrations Catalog

The ~80-connector plan (Klaviyo, Shopify, CallRail…) shipped in demand-ranked waves on a shared framework, so no client is ever blocked.

somedayL
PRD-10Social Hub

Compose, approve, schedule, publish socials from the command center — tied to actual site visits and CRM leads, not just likes.

somedayL
PRD-13Journeys & Funnels

Draw the ideal customer flow, overlay what visitors actually did — leaks become quantified findings. Its event-schema groundwork ships with v1.

Phase 0 = v1rest laterXL
03 · The merged plan

Two tracks, one bridge

Track A ships Atrium in weeks — the HOCATT team feels the consolidation by end of July. Track B builds AI Studio in months on Ed's roadmap, entered only after our PRD review trims it 60/40. The bridge makes Track A the instance-zero of Track B — nothing we ship now is throwaway.

July
August
September
Q4 →
TRACK A
ATRIUM
A0Fund + prep
A1The wrapper
A2Consolidation
A3Heavy migrations
TRACK B
AI STUDIO
B0PRD review + scope
B1Foundations
B2Core modules, cut 60/40
B3Fleet machinery
Jul 31 — one login, one portal, tickets live (tell Andre) Aug 6 — Jason's surgery · August runs without him M5 — HOCATT-equivalent command center rebuilt from adapters alone = the two tracks merge
BRIDGE RULE 1

Absorb, don't rewrite

Atrium's shell calls the proven backends — Faith, onboarding, GHL sync, control-center — as services. No duplicated builds.

BRIDGE RULE 2

One design system, one role model

Ed's tokens + Atrium's role/segment navigation. Decisions made for Atrium land as PRD feedback the same week.

BRIDGE RULE 3

Thin or reusable — nothing between

Everything in Atrium is either a disposably thin link/embed, or built clean enough to back out into the deployable base.

BRIDGE RULE 4

Real usage feeds the cuts

What the HOCATT team actually uses (and ignores) is evidence for the 60/40 PRD decisions — before Track B builds the equivalent.

BRIDGE RULE 5

Phase gates, not promises

Funding and team reality set the pace. Every phase has an explicit gate; no calendar heroics.

04 · Decision gates

Five gates keep both tracks honest

Each one is a real decision with an owner and a date — not a status meeting.

G1THIS WEEK

Green-light. Invoice sent, Ed looped on security fixes, greenfield-shell guardrail signed off.

G2~JUL 25

Atrium demo to Daniella; go / no-go on the end-of-July announcement to Andre.

G3END OF JULY

Scope freeze. PRD feedback consolidated, all-three session done, v1 base system defined — then design.

G4~AUG 1

Continuity plan. Every August work item has a non-Jason owner before surgery on Aug 6.

G5POST-M5

The switch. HOCATT migrates from wrapped systems to native AI Studio modules — only when parity is proven.

05 · How we review

Five questions for every PRD

Carry these into every page of the review. If a feature can't answer them, it's a someday.

Who's the buyer, and what problem does this uniquely solve?Filter and frame: micro-problem → macro-problem (money in, cost down, time saved, pain reduced).

Would any of the first five clients notice if this shipped six months later?The 60/40 test. If not — defer it.

What does this cost per client, per month?Isolation multiplies every infra, API, and AI cost by client count.

Does the simple version exist?The spec defaults to enterprise-grade. Ask what "startup mode" looks like.

Does HOCATT already have a working version?If yes: port or absorb it — don't respec it.

06 · Gaps & tensions

Four things the PRD set doesn't cover yet

Spotted while cross-referencing the PRDs against this week's meeting decisions. These are feedback for Ed, not criticism — the set is otherwise remarkably complete.

No ticketing PRD

A “baby ticket system” is a pre-launch requirement for Atrium (Faith needs somewhere to file support items) — and every future client will want one. Candidate PRD-14.

No customer-portal PRD

The my.hocatt.com concept — onboarding, training, shipment tracking, chat agent for customers — was adopted this week and generalizes to every client. Candidate PRD-15.

Two role models

PRD-02 fixes 4 roles; Atrium runs 9 roles + segments (“SEO is a view, not a role”). One answer needed before either build hardens.

Onboarding's home

HOCATT's onboarding system exists and ports in as-is — but the PRD set has no module for a client-business onboarding/ops system. Decide where it lives.