Ecommerce · Ireland / UK
An AI profit-analytics platform that tells online stores what to do next
Real-time profit across every sales and ad channel — with an AI analyst on top.
The challenge
Online retailers had sales, ad-spend and accounting data scattered across a dozen platforms, and no trustworthy view of true profit — let alone what to change.
What we did
We embedded a senior squad that built the data backbone — unifying 15+ commerce, advertising, payment and accounting systems into one contribution-margin model — then layered analytics and a conversational AI on top. We own delivery across backend, data engineering and ML.
The AI work
We built the AI layer: a governed data-access layer that lets LLMs query live store data safely; natural-language analytics ("why did profit drop last week?"); and always-on anomaly-detection agents that flag margin and spend issues in plain English with a learning feedback loop. Predictive models (customer lifetime value, churn) drive the recommendations.
Outcome
Merchants moved from static dashboards to asking questions in plain English and acting on AI recommendations; one new offer structure cut first-order acquisition cost by a double-digit percentage.
- Node
- Python (pandas, scikit-learn, Prophet)
- PostgreSQL / TimescaleDB
- Cube.js
- LLMs
- Redis / Celery