INSIGHTSBOT
InsightsBot is a GenAI-powered insights assistant purpose-built to automate End-of-Campaign (EOC) narratives and unlock proactive, data-driven recommendations. It connects campaign performance and operational context to generate market-ready insights that sales and insights teams can ship fast.
Engineered on Google Cloud with agentic orchestration, Text to SQL, and governed schema awareness, InsightsBot turns natural language questions into verified BigQuery SQL, validated results, and persuasive narratives. The result is a dramatic reduction in manual report writing and faster time-to-insight for teams.
"EdgeFirm transformed our EOC workflow with InsightsBot delivering an agentic architecture that blends Text to SQL, schema intelligence, and real time validation. Our teams now produce consistent, defensible insights in a fraction of the time, while proactively surfacing new opportunities. InsightsBot feels less like a tool and more like an insights teammate."
AUTOMATED END OF CAMPAIGN INSIGHTS THAT SELL THE NEXT CAMPAIGN
SCOPE OF WORK
Design and deploy a production-ready GenAI insights bot that automates EOC ‘Learnings & Recommendations’ and enables proactive, pre-campaign analysis through a natural language interface. The scope included agentic orchestration, Text to SQL, schema & metadata governance, Snowflake GCP data movement, authenticated UI, and analytics validation.
Key Deliverables:
- Agentic architecture (Router, Intent, Table, Column, SQL, Data Expert, Visualization agents)
- Text to SQL with validation and governed templates for BigQuery
- Schema/metadata extraction, enrichment, and versioning for robust SQL generation
- Snowflake GCP pipeline for curated analytics tables
- Secure, authenticated UI with WebSocket streaming and Markdown responses
- EOC narrative generation and recommendations aligned to KPIs and OKRs
DEVELOPED PIPELINES
Text-to-SQL & Validation: Natural-language questions are converted to BigQuery SQL via Gemini-powered prompts, validated for syntactic and logical correctness, and executed with traceability for defensible outputs.
TECHNOLOGY STACK
- Compute: Google Cloud Platform (GCP) for scalable agent and analytics workloads
- Storage: GCS for artifacts and intermediate data; governed datasets in BigQuery
- Database: BigQuery for analytics; Cloud SQL (Postgres + pgvector) for use-case/metadata embeddings
- LLM: Vertex AI (Gemini) for intent, SQL generation/validation, and narrative synthesis
- Backend Stack: FastAPI with LangChain and LangGraph for agent orchestration and tools
- Frontend & Auth: Next.js 14, WebSockets for streaming responses, NextAuth.js for secure access
- BI & Visualization: Config generation for UI charts; optional handoff to BI (e.g., Looker) for advanced visuals
RESULTS
- Manual Effort Reduced: Significant reduction in time spent drafting EOC ‘Learnings & Recommendations’ freeing teams to focus on QBRs and strategy.
- Speed to Market: Natural-language to validated SQL to polished narrative in minutes, not days accelerating advertiser engagement cycles.
- Consistency & Defensibility: Governed schema, verified queries, and dataset provenance improve accuracy and confidence in market facing materials.
- Pipeline for Growth: Framework generalizes to proactive, pre campaign analyses to surface net new opportunities.
- Business Alignment: Supports rebate and iROAS goals by scaling EOC quality, improving efficiency in under $5 categories, and tightening purchase cycles.
Conclusion
InsightsBot operationalizes GenAI for retail media analytics automating EOC narratives while unlocking proactive insights that convert into next campaign momentum. With agentic orchestration, governed Text to SQL, and authenticated self service, the platform delivers fast, defensible, and business aligned outputs that scale with data and demand.
Industry
Advertising Analytics
Built for measurement and reporting teams operating at enterprise scale, InsightsBot centralizes campaign data, customer signals, and DAP context into a self serve insights layer.
Headquarters
United States
OTHER PROJECT CASE STUDIES
EONA
Eona.ae, a dynamic brand serving the UAE market, sought to enhance its customer engagement and delivery operations through a conversational AI solution.
SOKRATEQUE.AI
AI-powered personal research assistant tailored for Master's and PhD students using RAG techniques and OpenAI API.
INPRO AI
Empower everyday people with the knowledge they need to understand their legal situations.
Ready to start a Project?
Share the details of your project with us, and our team will provide a tailored response aligned with your requirements. For direct communication, please reach out to us at email hello@edgefirm.io.