AI-driven · AEO/GEO-first · Closed-loop

Marketing as a Service

Point it at a company URL. It scans the site, writes the plan, generates on-brand creatives and landing pages, launches to every major ad platform — then optimizes spend on its own.

AEO/GEO-first · SEO second EmilBot onboarding Imagen 4 + Veo 3.1 creative Meta · Google · YouTube · TikTok · Snap Agentic autopilot
The problem

Performance marketing is too much work for too few people.

🧩

Fragmented stack

Strategy, creative, landing pages, six ad platforms, analytics and attribution — each a different tool, login and specialist.

Slow & manual

Briefs, design rounds, campaign setup and weekly optimization eat days. SMBs and agencies can't staff it.

📉

Spend leaks

Budgets sit on losers, winners get starved, and multi-touch credit is guessed. Nobody re-tunes daily.

The expertise to run it well exists — it just doesn't scale across hundreds of clients.

The solution

One agent that runs the entire marketing loop.

MaaS replaces the stack — and the busywork — with an AI operator that plans, creates, launches and continuously optimizes campaigns. A human approves; the agent does the rest.

Scan → Plan Idea → Creative Build → Launch Measure → Optimize

🧠 Strategy by Claude

Goals → strategy → funnel split, grounded in a live scan of the brand and its competitors.

🎨 Real generative creative

On-brand images via Imagen 4, video via Veo 3.1 — not stock, not templates.

🚀 First-class ad adapters

Launch real campaigns to Meta, Google, YouTube, TikTok and Snapchat.

The strategic bet

Built AEO/GEO-first — SEO second

Search is moving from ten blue links to one generated answer. MaaS is built to be the source the answer engines cite — ChatGPT, Perplexity, Gemini and Google AI Overviews — not just a ranking in a list that fewer and fewer people click.

💬

Answer-engine first

Every idea and page is optimized to be cited by the generative engines. Citation, not position, is the new top of funnel.

🏷️

Structured data by default

JSON-LD is wired into every generated page's <head> from the start — not bolted on afterwards.

📡

Weekly citation probes

Measures real domain citations (strong) and entity mentions (weak) across Perplexity / OpenAI / Gemini, caching the hit-rate per idea.

🔗

SEO as the secondary layer

Classic search stays — sitemap submission to Search Console + Bing — but sits beneath AEO/GEO in priority, not above it.

3 answer engines probed weekly JSON-LD on 100% of generated pages Priority: AEO/GEO > SEO
How it works

The closed loop

01

Scan

Crawl the site: brand color, palette, copy, competitors.

02

Plan

Claude drafts goals, strategy, funnel split & workflows.

03

Create

Ideas → multi-platform creatives + landing pages.

04

Launch

Real campaigns pushed to every connected platform.

05

Optimize

Daily insights → recommendations → autonomous action.

Every applied action snapshots baseline + post-action metrics — the loop literally learns which moves win, per client.

What's built today

A full marketing operating system

🤖

EmilBot wizard

URL → brand extraction → competitor research → plan → activated workflows.

💡

Idea → campaign

Angle, emotion and funnel stage per idea, each with multi-platform creatives.

🌐

Landing builder

Visual section editor, AI copy & hero images, Vercel deploys, custom domains.

🛒

Merchant Center

Auto-syncing datafeeds; pages auto-built from products; Shopping & PMax.

📊

Performance loop

Daily cross-platform insights with one-click apply on Claude's recommendations.

🔎

AEO/GEO engine · CORE

Probes Perplexity / OpenAI / Gemini for brand citations + JSON-LD action plan. Built in first, not bolted on.

📱

Mobile app

Expo / React Native iOS app — approvals, alerts, Live Activities, Siri intents.

⚙️

Workflow engine

GitHub-Actions-style automation with approval gates and run history.

The moat

It doesn't just suggest. It acts — and learns.

Most "AI marketing" tools stop at a draft. MaaS closes the loop with a calibrated autopilot that earns the right to act autonomously.

  • Learning loop — per-rule win-rate & lift materialized daily into effectiveness scores.
  • Budget orchestrator — reallocates daily spend across ad sets on trailing 14-day ROAS, bounded ±50%.
  • Calibration gates — shadow → assisted → auto, per client. Trust is configurable.
  • Markov attribution — data-driven multi-touch credit upstream from CAPI, not last-click.
±50%
bounded daily budget reallocation per run
3
trust modes — shadow · assisted · auto
14d
trailing ROAS window driving bids & budgets
Coverage

Every channel that matters, end-to-end

📘 Meta

Ads + Instagram organic. Campaign & creative creation wired live.

🔍 Google Ads

Search · Display · Shopping · Performance Max — fully wired.

▶️ YouTube

Video + Demand Gen, reusing the Google connection.

🎵 TikTok

Campaign + creative path for feed surfaces.

👻 Snapchat

WEB_CONVERSION fully wired — ad squad, media, creative, ads.

🔗 Search & feeds

Search Console + Bing sitemap submission; GMC product feeds.

Each integration ships a clear "env-missing" state until configured — demo-able from day one, productionizable when the keys land.

Under the hood

Built on a modern, durable stack

Application

Next.js 16 · App Router React 19 Tailwind 4 TypeScript Expo / React Native

Data & jobs

Drizzle ORM Neon Postgres (~40 tables) Inngest durable queue Vercel Blob

AI

Anthropic Claude + prompt caching Google Imagen 4 Veo 3.1 video MCP tool use

Security & tenancy

Multi-agency auth HMAC sessions + TOTP 2FA AES-256-GCM token vault Per-tenant authorization
Business model

Margins you can actually see

Every metered call — Claude tokens, embeddings, Imagen, Veo, AEO probes — records a usage event with provider, model, unit counts and microcents.

  • Per-client COGS — operating cost attributable to each client in real time.
  • Operator dashboard — 24h / 7d / 30d cost by provider, model, operation & client.
  • Agency multi-tenant — one wizard parks OAuth grants, each becomes its own client.

SaaS subscription + spend-based pricing, with transparent cost metering baked into the product — not bolted on.

5+
ad platforms launched from one console
~110
API route handlers across the product
100%
of AI spend metered to microcents
Where we are

Real product, demo-able today

✓ Shipping

  • EmilBot → plan → creatives → recommendations, end-to-end
  • Google Search/Display/Shopping/PMax live
  • Meta & Snapchat (WEB_CONVERSION) wired
  • Landing pages deploy to Vercel with custom domains
  • Autopilot learning loop + cost metering + mobile app

⟳ In flight

  • TikTok non-feed surfaces & catalog objectives
  • YouTube Target-CPV live smoke-testing
  • Meta catalog/audience targeting v2 schema
  • Consent-aware conversion tracking

Seed one command (pnpm seed) and walk the full loop on synthetic data — no platform keys required.

The vision

A marketing department
that runs itself.

From a single URL to a continuously-optimizing, multi-channel growth engine — built to be the answer the AI gives, not just a link in a list. At the cost of software, not headcount.

ScanPlanCreate LaunchOptimizeRepeat — autonomously

MaaS — Marketing as a Service