Project Status — Report #002

Post Brendon sync (May 28). Scope refined. Priority order updated. Two strategic paths identified.

57h

Human Hours (All Phases)

~$3.3K

AI Inference

12–16wk

Timeline

<2mo

Payback

📋 Changes from Brendon Sync (May 28)

Shopify context added: Retail = Shopify. Wholesale = manual (email/text). Potential future Shopify ↔ portal stock sync — not committed.
Prices corrected: Portal prices were Shopify retail. Brendon sent wholesale price list — ingest pending.
QB integration deferred: Brendon wants portal tested with real data first. QB moved from P0 to post-UAT decision.
Bulk customer import added: Brendon will export QB customer list (wholesale only) as Excel. Victor bulk-imports to avoid duplicates. New story added.
Pre-sprint data ingestion phase added: Products + customers must be loaded before Sprint 1 UAT begins.
Two strategic paths identified: Path A (AI email parsing) vs Path B (portal adoption). Decision pending after data ingestion + UAT.

Bottom Line

Portal is built. Real data ingestion is the immediate unlock. After UAT with a few trusted clients, Victor and Brendon decide: AI email intake first, or portal adoption first.

At a Glance

Portal Build

Live (Staging)

Admin + Customer views working

Data Ingestion

In Progress

Price list received; customer list pending

UAT

Blocked

Waiting on complete data

QB Integration

Deferred

Post-UAT decision

Strategic Path

Pending

A vs B — decide after UAT

Budget

On Track

No scope creep

Immediate Action Items

Brendon sent wholesale price list — Victor ingests into portal (product catalog update)
Brendon exports QB customer list (wholesale only, Excel) — target: next week
Victor bulk-imports customers after Excel received
Brendon receives admin login + portal link for UAT / stress testing
Victor + Brendon sync next week — review ingested data, decide Path A vs B

Context: Shopify + Wholesale Split

Channel	System	Process	Portal Role
Retail	Shopify	Automated	No change — Shopify handles retail
Wholesale	Manual (email/text)	~24h/week manual entry	This portal replaces manual process
Shopify sync	TBD	—	Possible future: Shopify ↔ portal stock sync (not committed)

Next Decision Point

After data ingestion + UAT

Victor and Brendon decide between two strategic paths. See Strategic Fork tab for full breakdown. This decision drives Phase 2 scope.

Phase Roadmap (Updated)

Phase	Scope	Status	Change
Pre-Sprint	Wholesale price list + customer list ingestion	In Progress	🆕 New — not in original spec
Phase 1 — MVP	Portal live, UAT with trusted clients	Ready after data	QB deferred to post-UAT
Phase 2 — Path A or B	AI email parsing OR portal adoption push — TBD	Pending decision	🆕 Fork introduced by Brendon sync
Phase 3	QB integration + remaining path	Not Started	Moved later by Brendon's request

Product Backlog — Updated

Post Brendon sync. QB reprioritized. Bulk import added. Strategic fork reflected.

P0 — Must Ship

P1 — Core Value

P2 — Complete MVP

P3/P4 — Future

What changed from v1

QB-01 / QB-02 downgraded from P0 → P1 — Brendon explicitly: test portal first, then QB
New story CM-08 added: Bulk customer import from CSV/Excel (QB export)
New Epic 0 added: Pre-Sprint data ingestion (price list + customers)
New Epic 9B added: Path A — AI email/SMS → order parsing (moved up from Phase 3)

🆕 Epic 0 — Pre-Sprint Data Ingestion

Portal is built. Real data must be loaded before UAT can start.

▾

ID	Story	Priority	Done When
DI-01	Ingest Brendon's wholesale price list into product catalog (correct prices)	P0	All products showing correct wholesale prices in portal
DI-02	Receive Brendon's QB customer export (wholesale only, Excel)	P0	File received from Brendon
DI-03	Bulk-import customers from Excel into portal (no duplicate risk)	P0	All wholesale customers visible in admin; no duplicates
DI-04	Share admin login + portal link with Brendon for UAT	P0	Brendon confirms access; can browse and test

Epic 1 — Foundation & Access Control

Everything depends on auth. No portal without this.

▾

ID	Story	Priority	Done When
AC-01	Owner logs in → full admin navigation visible	P0	Owner login → all menu items present
AC-02	Staff logs in with restricted access (no Settings, QB, Staff Mgmt)	P0	Staff login → limited nav; delete returns 403
AC-03	B2B client logs in → Customer Portal only	P0	Customer login → Customer Portal only
AC-04	Guest browses catalog without login	P0	Unauthenticated → catalog visible
AC-05	Brendon can deactivate a customer (access revoked, history preserved)	P1	Deactivated → "Account inactive" on next login
AC-06	No public self-registration — invite flow is the only path	P0	No sign-up page exists

Epic 2 — Product Catalog

Pre-requisite for all ordering flows.

▾

ID	Story	Priority	Done When
PC-01	Create product (name, desc, category, unit, price, availability, up to 10 images)	P0	Created → visible in customer catalog within 30s
PC-02	Set primary image and reorder via drag-and-drop	P1	Reorder → customer carousel reflects change
PC-03	Update product availability in real-time	P0	Set to 0 → "Out of Stock"; can't add to cart
PC-04	Set product status (Active / Inactive / Seasonal)	P1	Inactive → removed from customer view
PC-05	Soft-delete product (history preserved)	P1	Deleted → off catalog; past orders still reference it
PC-06	Manage categories (create, edit, delete)	P1	New category → available in product form immediately
PC-07	Image validation: max 5MB, JPEG/PNG/WebP only	P0	Wrong size/format → rejection message
PC-08	Price updates reflect in customer view within 60s	P1	Admin updates price → customer sees it on next load

Epic 3 — Customer Management & Onboarding Updated

Bulk import added from Brendon sync — avoids QB duplicates.

▾

ID	Story	Priority	Change
CM-01	Create customer profile (business name, contact, email, phone, address, schedule, tier, notes)	P0	—
CM-02	Send invite email to customer	P0	—
CM-03	Customer activates via invite link (pre-filled data, set password or magic link)	P0	—
CM-04	Invite links expire after 7 days (configurable)	P1	—
CM-05	Resend invite (new link, old link invalidated)	P1	—
CM-06	Convert guest order to registered customer (one click, data pre-filled)	P2	—
CM-07	Customer lifecycle: Invited → Active → Inactive	P1	—
CM-08	🆕 Bulk customer import from CSV/Excel (QB export format) — maps columns, deduplicates by email, creates accounts in batch	P0	🆕 Added from Brendon sync

Epic 4 — Ordering (Authenticated)

Core value. Eliminates 24h/week manual processing. No changes from v1.

▾

ID	Story	Priority
OR-01	Browse catalog with real-time availability (OOS disabled, low-stock warned)	P0
OR-02	Build order: select products, set quantities, see running total	P0
OR-03	"Repeat Last Order" — loads previous items, flags unavailable	P1
OR-04	Submit order → immediate confirmation (on-screen + email)	P0
OR-05	Admin sees new orders within 5 seconds	P0
OR-06	Admin confirms order (one click → packing sheet + QB invoice)	P0
OR-07	Admin adjusts qty before confirming → client notified automatically	P1
OR-08	Client cancels while New; can't cancel after Confirmed	P1
OR-09	Order status: New → Confirmed → Packed → Delivered (admin-only)	P0
OR-10	Availability decremented on confirmation, not submission (no phantom holds)	P0

Epic 5 — Guest Ordering

Acquisition funnel. No changes from v1.

▾

ID	Story	Priority
GO-01	Guest browses full catalog without login	P0
GO-02	Guest builds and submits order → PDF emailed	P0
GO-03	Guest order appears as "Pending Review (Guest)" in admin	P0
GO-04	Rate limit: max 5 guest orders per email per day	P2
GO-05	Convert guest to registered customer (one click)	P2

Epic 6 — Packing Sheets

Eliminates paper trail. No changes from v1.

▾

ID	Story	Priority
PS-01	Packing sheet auto-generated on order confirmation	P0
PS-02	Sheet includes: product name, qty, unit, bin/location, check-off boxes	P0
PS-03	Generate per-order or batched by delivery date	P1
PS-04	Optimized for print and tablet	P1

Epic 7 — Standing Orders

Eliminates 10+ clients worth of weekly re-entry. No changes from v1.

▾

ID	Story	Priority
SO-01	Create standing order template (client, products, quantities, schedule)	P1
SO-02	System auto-generates pending order 48h before delivery	P1
SO-03	Client notified by email to confirm or adjust	P1
SO-04	Client confirms with one click or modifies before deadline	P1
SO-05	Auto-confirm after deadline (configurable)	P2
SO-06	OOS product in standing order → flagged + client notified	P1
SO-07	Client and admin can edit template anytime	P2

Epic 8 — QuickBooks Integration Reprioritized

QB-01/02 downgraded P0→P1. Brendon: test portal first, then QB.

▾

ID	Story	Priority	Change
QB-01	Connect QuickBooks via OAuth 2.0	P1	Was P0 → downgraded. Post-UAT.
QB-02	QB invoice auto-created on order confirmation	P1	Was P0 → downgraded. Post-UAT.
QB-03	QB invoice cross-referenced with portal order number	P1	—
QB-04	Access token auto-refreshed	P1	—
QB-05	QB error → admin notified; order still processed	P2	—
QB-06	🆕 Customer sync QB ↔ portal — when customer added to portal, sync to QB to avoid duplicates (Brendon explicitly requested)	P1	🆕 Added from Brendon sync

🆕 Epic 9A — Path A: AI Email/SMS → Order Parsing

Removes bottleneck WITHOUT requiring clients to change behavior. Victor: "If this works, customers don't need the portal."

▾

ID	Story	Priority	Notes
AI-01	AI parses inbound order emails and maps to product catalog	P3	Path A. Decision pending post-UAT.
AI-02	AI parses inbound SMS/text orders	P3	Brendon: some clients text orders
AI-03	Ambiguous items flagged for human review	P3	—
AI-04	Parsed order shown as draft → admin confirms/edits → enters pipeline	P3	—
AI-05	Confidence scoring per line item (high/medium/low)	P3	—
AI-06	Confirmed parsed order → QB invoice auto-created	P3	Completes the email→QB chain

Epic 9B — Path B: Portal Adoption + Phase 2 Self-Service

Client-facing enhancements post-Phase 1 UAT.

▾

ID	Story	Priority
P2-01	Customer self-service account management	P3
P2-02	Order analytics dashboard for Brendon	P3
P2-03	Delivery tracking (status visible to client)	P3
P2-04	Mobile optimization	P3
P2-05	Shopify ↔ portal stock sync (TBD — mentioned but not committed)	P4

Sprint 1 Backlog — Updated

Pre-sprint data ingestion added. QB removed from Sprint 1 scope per Brendon's direction.

Pre

Data Ingestion First

Sprint 1 Stories

2wk

Duration

UAT

Exit Criteria

Sprint Goal

"By end of Sprint 1, Brendon can log in with real data, manage his catalog, invite real clients, and run UAT with a few trusted customers."

🆕 Pre-Sprint — Data Ingestion

Must complete before Sprint 1 UAT · Victor executes

The portal is built but has wrong data. Real prices + real customers must be loaded before Brendon can meaningfully test anything. This unblocks Sprint 1 UAT.

Task	Input	Done When
DI-01 — Ingest price list	Wholesale price list (received from Brendon)	All products showing correct wholesale prices
DI-02 — Receive customer export	Brendon's QB export (wholesale clients only, Excel)	File received
DI-03 — Bulk import customers	Excel from DI-02	All wholesale customers in admin; no duplicates
DI-04 — Share admin access	—	Brendon confirms he can log in and stress-test

Track A — Infrastructure & Auth

Est: 3 days · AC-01 to AC-06

Foundation for everything. Build first, in full.

Task	What It Means	Done When
A1 — Supabase Setup	Create project, configure auth, define 4 roles: owner, staff, customer, guest	4 roles working; local env connects
A2 — Cloudflare Scaffold	Init frontend repo, deploy to staging	staging.marisgardens.com returns 200
A3 — Owner Login	Email + password; redirect to Admin Portal	Owner logs in → full nav
A4 — Staff Login	Restricted nav (no Settings, QB, Staff Mgmt)	Staff login → limited nav; delete returns 403
A5 — Customer Login	Redirect to Customer Portal; admin URL blocked	Customer login → portal only
A6 — Route Guards	All routes protected per role	Unauthenticated /admin → login redirect

Track B — Product Catalog

Est: 4 days · PC-01 to PC-08

Build the full CRUD + image pipeline so Sprint 2 can focus entirely on ordering. Prices must reflect Brendon's wholesale list (not Shopify retail).

Task	What It Means	Done When
B1 — Product Data Model	Supabase table: products (id, name, desc, category_id, unit, price, availability, status)	Migration runs clean
B2 — Image Storage	Supabase Storage; max 5MB; JPEG/PNG/WebP only	Upload validates type + size
B3 — Image Ordering	sort_order join table; drag-and-drop UI	Reorder → persisted; carousel reflects
B4 — Category CRUD	categories table; admin CRUD	Create/edit/delete → available in product form
B5 — Product CRUD	Create, edit, soft-delete; status toggle	Full CRUD; soft delete preserves history
B6 — Real-time Availability	Admin update → customer sees <60s	Set to 0 → OOS shown; can't add to cart
B7 — Customer Catalog View	Product grid; filter by category; search	Customer sees catalog; OOS disabled

Track C — Customer Management & Onboarding

Est: 3 days + bulk import · CM-01 to CM-08

Now includes bulk import (CM-08) from Brendon's QB export. Build invite flow + bulk import in Sprint 1 so data ingestion can happen immediately after.

Task	What It Means	Done When
C1 — Customer Data Model	Supabase table: customers (all fields)	Migration runs clean
C2 — Admin Customer Creation	Form + Supabase auth invite	Profile saved; visible in list
C3 — Invite Email Flow	Resend integration; branded email; 7-day link	Email within 60s; link works
C4 — Customer Activation	Pre-filled form → confirm → account active	Activation completes end-to-end
C5 — Invite Expiry + Resend	Day 8 → error page; resend invalidates old link	Both cases handled
C6 — 🆕 Bulk CSV/Excel Import	Upload QB export → map columns → deduplicate by email → batch create accounts	Brendon's QB export imported cleanly; no duplicates

Definition of Done — Sprint 1

All 3 tracks deployed to staging
Role matrix fully enforced (owner/staff/customer/guest)
Product catalog with real wholesale prices working in staging
Customer invite flow + bulk import working end-to-end
Brendon's real customer list imported with no duplicates
Brendon completes UAT with at least 2 trusted clients placing test orders
No QB integration in Sprint 1 — post-UAT decision only

What's NOT in Sprint 1 (updated)

Feature	When	Why Deferred
QB integration	Post-UAT	Brendon explicitly: test portal first, then QB
AI email/SMS parsing	Path A decision	Depends on UAT + strategic fork decision
Ordering flows	Sprint 2	Depends on auth + catalog (this sprint)
Packing sheets	Sprint 3	Depends on ordering (Sprint 2)
Standing orders	Sprint 3	Depends on ordering (Sprint 2)
Shopify sync	TBD / Phase 3+	Mentioned by Victor — not committed by Brendon

Strategic Fork

After data ingestion + UAT, Victor and Brendon choose one path. Each solves the core problem differently.

The Core Problem

Brendon's team spends ~24h/week manually processing orders received by email and text. Every restaurant has a different format. This is the bottleneck — and there are two ways to remove it.

The Two Paths

Path A — Remove the Bottleneck

AI Email + SMS → Order Pipeline

Build AI that reads inbound emails and texts, maps them to the product catalog, and creates orders automatically. Restaurants keep doing exactly what they do today.

AI parses "3 cases romaine, 2 bags herbs" from email
Maps to product catalog; flags ambiguous items
Draft order presented to Brendon → one-click confirm
Confirmed order → QB invoice auto-created
SMS channel supported (some clients text)

"If this works, you don't need your customers to use the portal." — Victor, May 28 sync

Path B — Change Client Behavior

Portal Adoption Push

Invest in getting the 70+ B2B clients to actively use the ordering portal instead of email/text. Higher long-term payoff, harder adoption curve.

Client onboarding + training materials
Mobile optimization (orders from the field)
Customer self-service (address, preferences)
Analytics dashboard for Brendon
Delivery tracking visible to clients

Higher ceiling — clients who use the portal generate cleaner data, less ambiguity, faster processing. But adoption takes time and trust.

Path Comparison

Dimension	Path A — AI Email	Path B — Portal Adoption
Client behavior change	None — zero friction	High — requires onboarding + habit shift
Time to value	Fast — works with existing email/text	Slow — depends on client adoption rate
Data quality	Medium — AI can misparse ambiguous requests	High — structured input, no ambiguity
QB integration dependency	Required for full chain	Optional (enhances but not required)
Effort estimate	~10h + ~$650 inference	~12h + ~$650 inference
Risk	AI parsing accuracy — wrong fills	Client resistance — low adoption
These paths exclusive?	No — both can be built. The decision is which to do first.

Recommendation

Victor's framing from the sync

Path A removes the bottleneck immediately and requires no behavior change from clients. If AI parsing accuracy is high enough, Brendon sees ROI before a single client logs into the portal. Path B is the long-term play — cleaner data, higher engagement. The strongest outcome: ship Path A first to prove ROI, then use that credibility to drive Path B adoption.

Decision gate — after UAT

Victor and Brendon sync after data ingestion + UAT. Brendon's experience using the admin portal with real data will inform which pain point feels most urgent to solve next.