De Generals

StrategIQ is an AI-powered platform that automates marketing creation, localization and sentiment analysis—helping brands deliver authentic, data-driven, culturally resonant campaigns faster, smarter.

Comment

Inspiration

We were inspired by how difficult it's become for brands—especially small businesses—to keep up with the constant demand for engaging, culturally relevant marketing content. Many waste hours creating campaigns that fail to resonate globally or adapt to cultural nuances.

What it does

StrategIQ is a unified AI platform that automatically generates, localizes, and optimizes campaigns using real-time market intelligence. Our goal was to eliminate repetitive manual work and empower teams to create data-driven, audience-focused marketing effortlessly.

How we built it

We built StrategIQ as a fully serverless, event-driven platform on AWS. Our backend leverages Python-based AWS Lambda functions for all compute, orchestrated via API Gateway and EventBridge Pipes. We designed an agentic architecture using Amazon Bedrock Agents, which coordinate specialized Lambda action groups for image analysis, data enrichment, and cultural intelligence. Product images and assets are stored in Amazon S3, while campaign data and metadata are managed in DynamoDB. For AI-powered insights, we integrated Amazon Rekognition for image analysis, Bedrock (Claude/Nova) for campaign generation and localization, and AWS Comprehend for sentiment analysis. Real-time market intelligence is sourced from the YouTube Data API. Infrastructure is provisioned and managed using Terraform for reproducibility and scalability. The frontend is built with React and Vite, enabling users to securely upload assets via presigned URLs and interact with the campaign generation workflow seamlessly.

Challenges we ran into

  • Action Group Orchestration: Ensuring Bedrock Agents correctly invoked the right Lambda functions in the proper order.
  • Data Synchronization: Managing DynamoDB streams and EventBridge Pipes without data duplication or missed triggers.
  • External API Quotas: Handling YouTube Data API rate limits and building graceful fallback mechanisms.
  • Prompt Optimization: Maintaining model token efficiency and coherent outputs across diverse campaign data.
  • Latency: Balancing real-time generation speed with the depth of campaign insight.

Accomplishments that we're proud of

  • Successfully built a unified, agentic AI platform that automates campaign generation, localization, and optimization for global audiences.
  • Seamlessly integrated multiple AWS services—Bedrock Agents, Lambda, Rekognition, DynamoDB, EventBridge, and Comprehend—into a robust, event-driven pipeline.
  • Achieved secure, credential-free asset uploads using presigned URLs and S3.
  • Delivered real-time market intelligence and cultural adaptation through specialized action groups.
  • Maintained schema consistency and high output quality across diverse campaign scenarios.
  • Enabled rapid, scalable deployment and reproducibility using Terraform.
  • Created an intuitive frontend experience with React and Vite, empowering users to launch campaigns effortlessly.

What we learned

Through this project, we deepened our understanding of AWS Bedrock Agents, EventBridge Pipes, and serverless orchestration patterns. We learned how to combine multiple AWS services—Rekognition, DynamoDB, Lambda, and Bedrock—into a fully event-driven AI pipeline. We also discovered the importance of schema consistency across Bedrock action groups, prompt optimization for generative models, and using Terraform for reproducible infrastructure deployment.

What's next for De Generals

We're focused on expanding StrategIQ with multi-agent orchestration, enabling custom-built agents that are fine-tuned for specific workflows and verticals. We're integrating advanced machine learning for accurate forecasting and predictive analytics, and planning to give agents secure, permissible access to product inventory for dynamic campaign adjustments. Future enhancements include multi-platform trend sources (TikTok, Instagram), real-time campaign performance feedback, automated variant testing, and workspace segmentation for multi-tenant support. Our goal is to deliver even smarter, more adaptive marketing automation for global brands and small businesses alike.

Built With

Share this project:

Updates