Build a closed-loop autonomous attribution agent connecting top-of-funnel ad data to bottom-of-funnel CRM lead quality — then autonomously reallocating Google and Meta budgets toward high-quality leads instead of cheap clicks.
Results
Live Demo: agent.rukesh.in
Attribution: Closed-Loop
Deployment: Self-Hosted
Built In: Hackathon
Methodology
Map
Connect
Score
Route
Deploy
Strategy & Execution
Attribution Architecture: Designed a closed-loop system connecting Google Ads and Meta ad signals to CRM lead quality scores — tracking the full journey from ad click through to lead outcome and revenue signal.
n8n + Python Pipeline: Built the core pipeline in n8n with Python scripts handling lead quality scoring, signal normalisation, and budget reallocation logic across platforms.
OpenAI Codex Integration: Used OpenAI Codex to automate budget routing decisions — the agent analyses lead quality patterns and surfaces reallocation recommendations without manual intervention.
Self-Hosted Deployment: Deployed on Docker and Coolify on a self-hosted VPS — live demo accessible at agent.rukesh.in. Designed for full reproducibility and zero cloud dependency.
What I Delivered
Attribution Engine: Closed-loop pipeline connecting ad spend data to CRM lead quality signals across Google and Meta.
n8n Automation Workflow: Multi-node pipeline for data ingestion, lead quality scoring, and autonomous budget reallocation.
Budget Router: AI-driven logic reallocating spend toward high-quality lead sources without manual input.
Live Demo: Self-hosted on Docker/Coolify — accessible at agent.rukesh.in.