MarAI – AI-Native Multi-Client Marketing Intelligence & Content Automation Platform_

Project Overview

Modern marketing teams do not need isolated content generators.

They need:

• Multi-client orchestration
• AI-native content production
• Brand-context injection
• Citation visibility tracking
• SEO + GEO optimization
• Session-persistent intelligence
• Export-ready deployment
• Secure, revocable authentication
• Backend-authoritative modeling

Traditional marketing tools provide:

• Static content generators
• Disconnected dashboards
• Surface-level SEO audits
• Keyword visibility reports
• Campaign scheduling

They do not provide:

• AI-context memory across sessions
• Client-scoped content isolation
• Citation visibility scoring in AI systems
• Dual-model intelligence pipelines
• Retrieval-aware optimization
• Deterministic backend-controlled AI orchestration
• Token-aware LLM session management
• Structured multi-tool persistence

They do not operate as unified AI-native marketing infrastructure.

MarAI was built to solve that gap.

It is not a content generator. It is a full-stack AI marketing execution platform.

What It Does

MarAI enables agencies and marketing teams to:

• Manage multiple clients securely
• Inject brand context into every AI request
• Generate 12+ content types
• Track brand visibility in AI systems
• Optimize content for SEO + generative AI discovery
• Persist session intelligence across tools
• Export assets in PDF, Excel, and Word formats
• Store structured marketing assets
• Integrate with Google Analytics & Search Console
• Enforce backend authority across scoring & intelligence

It is a unified AI-native marketing operations system.

Core Capabilities

• Dual AI Intelligence Architecture

  Uses Google Gemini 2.5 Flash as the primary content generation engine for persona generation, content creation, SEO strategy, and context-aware rewriting with 50K token handling. Uses Perplexity Sonar Pro as the citation and brand visibility intelligence engine for brand mention detection, share-of-voice analysis, and prompt-based retrieval scans. All scans persist in PostgreSQL.

• Multi-Client Scoped Architecture

  Every output is linked to a specific client via client_id association, structured brand metadata, goals and budget tracking, brand guidelines injection, and context retention when switching clients. Ensures strict data isolation and secure multi-tenant design.

• Session Persistence Pattern

  Each tool maintains conversation history, token count, context URLs, generated output, and last analysis timestamp. Switching between tools does not destroy intelligence. State is managed centrally via App-level orchestration with no accidental intelligence loss.

• Integrated Marketing Suite (10 Modules)

  Covers real-time dashboard analytics, marketing/social/email calendars, AI content creator, persona builder, keyword research, content gap analyzer, SEO strategy generator, content optimizer (SEO/GEO), citations tool, and a saved asset library with 100-item capacity, type-based filtering, and one-click reload.

• SEO / GEO Content Optimizer

  Evaluates word count, heading structure, internal links, image usage, keyword density, informational coverage, AI discoverability (GEO score), and brand alignment score. Performs section-by-section analysis, competitor gap comparison, AI rewrite suggestions, and structured PDF/Word exports. Bridges SEO and generative AI optimization.

• Citation Intelligence Engine

  Allows brands to measure whether they are cited in AI responses, share-of-voice vs competitors, top ranking domains in responses, and prompt-specific visibility shifts. Models AI-powered retrieval exposure — not keyword ranking, but AI visibility modeling.

• Asset Intelligence Library

  Supports 10 structured asset types including persona, marketing calendar, social calendar, email calendar, content, SEO strategy, content gap, content optimizer, keyword research, and citations. Enforces 100-item limit per user with JSONB metadata indexing, fast retrieval, client-side preview, and export per asset type.

• Export Intelligence Engine

  Client-side generation of styled PDFs, structured Excel files, and Word documents. Maintains consistent branding, chart exports, citation reports, and SEO optimizer reports. Bridges AI intelligence with real-world deployment workflows.

• Enterprise Security & Authentication

  Implements opaque token-based sessions, max 10 active tokens per user, DB-validated session revocation, automatic token cleanup, password reset invalidation, rate limiting (100 req/15 min), Helmet.js security, SQL injection prevention, and GDPR-ready structure. Security enforced at database level.

The Challenge

Marketing teams face:

• Fragmented AI tools
• Loss of conversation memory
• No client-scoped context injection
• No citation visibility tracking
• No GEO optimization modeling
• No unified asset storage
• No secure multi-client orchestration
• No backend-controlled scoring authority

Traditional platforms separate:

• Content
• SEO
• Analytics
• Scheduling
• AI

They do not unify them.

The Solution

Built a full-stack AI-native marketing platform composed of:

Backend:

• Express + TypeScript API
• Gemini integration layer
• Perplexity citation service
• Content optimizer service
• Google OAuth service
• Token management service
• Client service
• Web scraping layer (Puppeteer + Cheerio)
• PostgreSQL persistence
• Migration-controlled schema
• Audit logging

Frontend:

• React 18
• TypeScript strict typing
• Session-persistent architecture
• Executive dark-mode dashboard
• Modular page routing
• Recharts data visualizations
• Export modal system
• Strict client-scoped tool orchestration

The system enforces backend authority — visibility scores, citation scans, and SEO metrics cannot be manipulated client-side.

Why It Matters

AI-native search is changing marketing behavior.

Brands must understand:

• Whether AI systems surface them
• Whether content is GEO optimized
• Whether sessions retain brand context
• Whether AI outputs align with brand guidelines
• Whether marketing workflows scale across clients
• Whether AI outputs are structured, stored, and deployable

MarAI provides structured AI marketing infrastructure — not isolated generation utilities.

It transforms marketing execution into AI-orchestrated, session-persistent, citation-aware, export-ready intelligence operations.

Future Expansion

• Workflow approvals system
• Slack / Asana integration
• Batch AI generation
• Competitive citation tracking
• Trend intelligence engine
• Scoring weight customization
• Persistent analytics history
• SaaS multi-organization layer
• AI memory optimization engine
• Executive PDF reporting suite
• Black / Orange premium dashboard redesign