Let’s Forage

Challenges

Let’s Forage recognized a critical, high-value gap in the advertising market: the inability for global brands to create highly personalized, culturally authentic ad campaigns at massive scale. The client required a robust engineering solution to transform its proprietary, locally-run AI data models into a production-grade, enterprise SaaS application that could:

1. Provide a secure, high-performance web interface for data visualization and ad configuration.
2. Automate a multi-step AI pipeline for creative content generation.
3. Orchestrate end-to-end campaign deployment, performance tracking, and optimization through the Meta advertising ecosystem.

Solutions

Zazmic partnered with Let’s Forage across two development phases to deliver a complete, scalable, end-to-end platform built on a zero-DevOps microservices architecture using Google Cloud Platform. The solution encompassed front-end development, backend service creation, AI workflow migration (MLOps), and deep integration with the Meta Ads platform.

Phase I: Foundation & Core Application

The initial phase focused on building the secure, user-facing application and migrating the client’s core data integration tasks into the platform’s backend.

• User & Access Control (RBAC): Full-stack development of user registration, login, and an Admin Panel (React/Node.js) to manage user roles, ensuring compliance and data security.
• Responsive UI/UX: Designed an intuitive web application optimized for modern browsers, transforming complex datasets into easily navigable insights.
• Meta Search Integration: Integrated backend APIs for Meta hashtag searches, storing raw data in BigQuery and initiating the client’s ML pipeline for core content analysis.
• VisionBoard & Dashboards (APIs): Developed functional backend APIs (Node.js) and front-end integration (React) for the VisionBoard and Insights Dashboards, serving data from BigQuery.
• Semantic Search & Retrieval: Implemented advanced semantic search capabilities on top of core data, leveraging vector embeddings to allow users to search for content based on themes and context rather than just keywords.

Phase II: AI-Powered Campaign Orchestration

During the second phase, Zazmic transitioned to building the core AI, Ad-Tech, and Data Engineering pipelines, migrating the client’s local AI workflows to a robust, scalable GCP architecture.

• End-to-End ETL Orchestration(MLOps): Migrated locally-run scripts to Cloud Composer DAGs (Apache Airflow).  This defined sequential, repeatable steps for data scraping, evaluation, and storage optimization.
• AI-Driven Creative Briefing: Utilized Gemini 1.5 to compile advertiser input with cultural social media trends into an “Ad-Set Brief” that informs creative strategy.
• Generative AI Pipeline: Integrated Imagen 3.0 to take text-to-image prompts from the Gemini brief and automatically generate high-quality, culturally relevant image variations for the campaign.
• Ad Campaign Publishing: Developed a Cloud Run service to seamlessly push the generated images, ad copy, and audience targeting specifications to the Meta Ads API, creating hundreds of customized ad sets at scale.
• Performance Feedback Loop: Established a Meta Ads API integration to ingest campaign performance data (impressions, clicks, cost-per-click) back into BigQuery. This data is used to continuously refine and optimize content generation AI models.
• Scalable Cloud Architecture: Established a zero-DevOps, microservices architecture using Cloud Run to host the Frontend, Backend, API Scrapers, and Ad Publishing services, ensuring rapid scaling and cost-effectiveness.

Outcomes

Zazmic successfully delivered a first-of-its-kind AI-powered AdTech orchestration platform, positioning Let’s Forage as a disruptive leader in the market:

• Global Reach: We successfully productionized and scaled the client’s proprietary ML pipeline to conduct deep, semantic analysis across various content in multiple languages.
• Efficiency: The delivered AI-to-Ad pipeline enables the client to generate thousands of market-tested ad creatives with efficiency far exceeding traditional agency models.
• Enterprise Scalability: The robust MLOps and microservices architecture ensures the platform can reliably handle massive data ingestion and high-volume campaign orchestration necessary for global enterprise clients.
• Maximized ROI: By closing the loop between ad performance data and content generation, the system maximizes ad relevance, leading to improved campaign ROI.