7 kits, from $14 — built on Supabase, tested in production
Production-Ready Supabase Modules.
Not Another Todo App.
Drop-in kits with real RLS policies, Edge Functions, TypeScript types, and migration files. Built on Supabase, tested in production.
Every Supabase tutorial teaches you the basics.
None ship the hard parts — proper RLS that doesn't lock you out in production, Edge Function patterns that handle errors gracefully, real-world schema design with actual indexes, and migration files that don't break on deploy. These kits do.
What you actually get
Not boilerplate. Production patterns.
Every kit ships with the files you'd write yourself if you had the time. Except these are already tested.
Real RLS Policies
Not auth.uid() = user_id and done. Multi-role policies, service-key bypasses, and the edge cases tutorials skip.
Edge Functions
Deno Edge Functions with proper error handling, CORS headers, typed request/response, and environment variable patterns.
Migration Files
Proper supabase/migrations/ with up/down, indexes, enums, and comments. Run supabase db push and go.
The kits
Pick one or grab the bundle.
Each kit is a self-contained module. Drop it into your existing Supabase project.
Invoice Kit
AI voice parsing to structured invoices with PDF generation. Speak an invoice, get a typed record and a downloadable PDF.
Receipt Kit
Photo OCR to categorization with bank CSV matching. Snap a receipt, match it to your bank export, categorize automatically.
Inventory Kit
Multi-tier pricing, real-time stock tracking, and voice logger. Watch inventory update live across tabs with Supabase Realtime.
Client Intelligence Kit
Health scoring, seasonal pattern detection, and churn alerts. Postgres functions + pg_cron that run inside your database.
PWA Notifications Kit
5-layer push notification fallback that actually works on iOS. Service workers, Notification API, in-app banners, and Supabase-backed delivery tracking.
Brain Kit
pgvector semantic memory with MCP protocol server. A Deno Edge Function that gives any AI tool persistent, searchable memory over your Supabase database.
Recipe Memory Kit
Voice and photo capture with multilingual transcription and AI tagging. Record a recipe in any language, get structured data back.
Under the hood
Code you'd actually ship.
Not simplified for a tutorial. Not stripped down for a demo. Real patterns from a production app.
-- Not just auth.uid() = user_id create policy "team_members_access" on invoices for select using ( auth.uid() = owner_id or auth.uid() in ( select user_id from team_members where team_id = invoices.team_id and role in ('admin', 'billing') ) );
import { serve } from "std/http/server.ts"; import { Invoice, ParseResult } from "./types.ts"; serve(async (req: Request) => { if (req.method === "OPTIONS") return corsResponse(); const { audio, locale } = await req.json(); const result = await parseVoice(audio, locale); if (!result.ok) return errorResponse(result.error, 422); return jsonResponse<ParseResult>(result); });
export interface Invoice { id: string; owner_id: string; team_id: string; client_name: string; line_items: LineItem[]; status: "draft" | "sent" | "paid" | "overdue"; total_cents: number; currency: string; due_date: string; created_at: string; pdf_path: string | null; }
the bundle
Dash Biz Suite
All 7 kits. One price. Every future update included.
- Invoice Kit
- Receipt Kit
- Inventory Kit
- Client Intelligence Kit
- PWA Notifications Kit
- Brain Kit
- Recipe Memory Kit
The stack
Built on Supabase. All the way down.
Every kit uses real Supabase primitives. No ORMs. No wrappers. Just Supabase.
From the blog
Deep dives on the patterns inside these kits.
Replacing 15 LaunchAgents with pg_cron and Edge Functions
How I migrated a production system from local cron jobs to fully serverless Supabase scheduling. The same patterns used in the Client Intelligence Kit.
Read post → Edge FunctionsBuilding Serverless Agents on Supabase Edge Functions
The Edge Function architecture behind the Brain Kit and Invoice Kit. Typed handlers, error boundaries, and CORS patterns that work in production.
Read post →Stop writing the same RLS policies from scratch