Pricing
Portfolio Case Study

The Whitmore Platform

A full-stack AI marketing platform built solo. Multi-tenant architecture, Claude-powered agent pipeline, and deep integrations with the tools local businesses already use. Running in production with real clients.

The Problem

Local businesses are stuck between too cheap and too expensive

Self-serve tools like Hootsuite, Mailchimp, and Canva cost under $100/mo but require constant manual effort. Full-service agencies cost $1,500–$3,000/mo but offer limited personalization and opaque reporting. The gap between "DIY with too many tabs open" and "pay someone to manage it" was wide open.

The specific insight: Claude and native MCP integrations could automate most of the agency workflow — if someone actually built the implementation layer, the multi-tenant infrastructure, and the client-facing UX. That's what the Whitmore platform is.

The Solution

A production AI marketing platform, built solo

The Whitmore platform is a multi-tenant SaaS application where each client organization gets an isolated Claude agent workspace connected to their actual marketing tools. The agent can create campaigns, schedule social posts, pull analytics, generate creatives in Canva, and respond to leads — all from a single chat interface.

The connector framework is the architectural bet: build each integration once (Meta, Google Ads, Canva, GA), and every client who activates that connector gets all the tools, skills, and dashboard views automatically. No per-client custom dev.

The whole platform was built by one engineer — no co-founder, no dev team, no outsourcing. It went from zero to production clients in under 60 days.

Tech Stack

The full stack

Next.js 15

App Router, RSC, streaming

Supabase

Postgres, RLS, realtime

Claude API

Agent pipeline, tool use

Meta Graph API

FB + IG posts, analytics

Google Ads API

OAuth, 10 agent tools

Canva Connect

Design gen + export

Google Analytics

Reporting + sync

Vercel

Edge deploy + analytics

What Was Built

Feature breakdown by phase

1

Foundation

  • Multi-tenant auth (org-scoped RLS on every table)
  • Claude agent pipeline with tool use and streaming
  • Connector framework — each connector brings tools, skills, and dashboard views
  • AES-256-GCM token encryption for OAuth credentials
2

Integrations

  • Meta / Instagram connector — post scheduling, analytics, comment webhooks
  • Google Ads connector — OAuth flow, 10 agent tools, reporting dashboard
  • Canva Connect — design generation, export, artifact storage
  • Google Analytics — site reporting with full sync
3

Campaign Engine

  • Campaign container with 12 asset roles and 5 agent tools
  • Carousel approval workflow (draft → approve → publish)
  • Artifact system — PDFs, CSVs, presentations as in-app cards
  • Anti-hallucination prompt guardrails
4

Production Hardening

  • RLS on 7 tables, storage bucket org-scoping, search sanitization
  • OAuth HMAC + AES-256-GCM for all connector tokens
  • 15 integration tests + eval architecture (unit, agent capability, browser)
  • Mobile-first UX — full-page chat, overlays, collapsible sections

Outcome

Running clients. Proven model.

Obsessed Closets (custom closet company) and AB Construction (general contractor) are both live on the platform. Social posting pipeline is automated. Campaign management is active. Analytics are reporting.

The platform is also the most direct proof of what we do as a consultancy: when a client needs a custom AI implementation, this is the quality and architecture they get.

The takeaway for prospective clients

"This is the kind of thing I can build — or configure in Claude, depending on what you actually need. The Whitmore platform exists because I built it. Your implementation will be right-sized for your business — not over-engineered, not under-built."

— Eugene Cyckowski, Whitmore AI

Want something like this for your business?

Start with the free AI audit. We'll figure out what's right for your scale and budget — whether that's a simple Claude setup or something more custom.

Or email: hello@heywhitmore.ai