What actually works when you put AI & analytics into production?
AI, automation & analytics: notes from production.
Finzarc is an AI and data-engineering studio; Insights is where we write down what we learned shipping the systems, not what we read about them. Every field note comes from a build that reached production: what AI, automation and analytics actually changed inside a real enterprise, and where most implementations quietly fail. New here? Start with the pillar note, AI and analytics in enterprises: what works in production.
Who writes Finzarc's field notes, and about what?
Insights is Finzarc's field-notes journal on putting AI, automation and analytics into production. We only publish notes tied to a delivered build, eighteen and counting, so what you read here is what survived contact with a real enterprise, not a prediction about one. The writing comes from founders Piyush Kumar and Abhinav Tripathi and the team that ships: the people you meet are the people who build. If you're new, start with the flagship essay on what actually works when you put AI and analytics into production. Everything else is grouped below by what it's actually about.
Why does enterprise analytics fail to change decisions?
Most enterprise analytics fails at the last mile: the moment a number is supposed to change a decision. These field notes live in that gap: dashboards that look like control but trigger no behaviour, the real-time-versus-batch call that quietly sets your architecture bill for years, and data platforms that shipped without ever delivering value. If you'd rather see it built than read about it, this is the analytics and business intelligence we ship.
Why dashboards create the illusion of control
Real-time analytics vs traditional BI: which one your business actually needs
How data platforms fail to deliver business value, and what to build instead
Does predictive analytics really improve inventory and demand planning?
Forecasting works. Implementations fail, and the difference is almost never the algorithm. These notes cover where predictive analytics actually moves inventory, demand planning and revenue, and the operational reasons most forecasts never reach a decision. For the delivered versions, see our optimal price-point analytics build and the revenue-maximisation platform we shipped for a consumer-goods major: +27% revenue on hair serum, margin 30% → 38%.
How predictive analytics improves inventory and demand planning
Where AI actually improves revenue in retail and FMCG
Why do AI pilots stall, and what ships instead?
Models are commoditising. Shipping is not. This cluster is about why enterprise AI stalls in pilots and staging, and why execution speed, idea to production in about three weeks, is becoming the real advantage. It's the same argument behind why teams choose Finzarc: working software over promises of future. Want it built instead of read about? See the AI agents and agentic automation we build.
Execution speed is becoming the real AI advantage
no notes match this filter.
One shipped lesson a week. Before it's public.
read past issues on substack ↗Reading is research. Shipping is the point.
See our delivered AI and analytics case studies, then take 30 minutes with the founding team: bring the problem, leave with a scope.
Piyush Kumar & Abhinav Tripathi: meet the founders who build and sell FOUNDERS · THE PEOPLE ON YOUR CALL Finzarc is an India-based AI and data-engineering studio that ships production automations, analytics, business intelligence and custom applications for enterprises (from consumer goods and FMCG to retail, manufacturing, industrial and financial services) with first delivery in about three weeks. Insights is our field-notes journal, written by founders Piyush Kumar and Abhinav Tripathi and updated with one shipped lesson at a time. Recurring themes: why AI pilots stall before production, business intelligence that actually changes behaviour, predictive analytics for inventory and demand planning, and real-time analytics versus traditional BI. Bring a problem worth shipping and book a 30-minute scoping call.
Frequently asked questions about Finzarc Insights
What does Finzarc write about in Insights?
Insights is Finzarc's field-notes journal on AI, automation and analytics in production. Each note comes from a delivered build: what actually worked inside a real enterprise, and where most implementations fail. Recurring topics: enterprise AI, business intelligence, predictive analytics, data engineering and custom applications.
Who writes Finzarc's field notes?
Founders Piyush Kumar and Abhinav Tripathi, and the team that ships the systems. Finzarc is a studio where the people you meet are the people who build, so the notes come from production, not from theory or a content desk.
How often is Insights updated?
Roughly one shipped lesson a week. It's published first to subscribers on Substack ('Field Notes, weekly, by Piyush') and then here. We only publish notes tied to a real build, so cadence follows delivery.
Where should I start?
Start with 'AI and analytics in enterprises: what works in production', the pillar essay distilling eighteen delivered systems. From there, branch into the analytics and business intelligence, predictive analytics and forecasting, or AI-in-production clusters.
Do these field notes apply to my industry?
Most draw on consumer goods and FMCG, retail, manufacturing and financial services, the sectors Finzarc ships into. The failure modes we write about (pilots that stall, dashboards that don't trigger action, forecasts that never reach a decision) are industry-agnostic.
How is this different from a typical AI blog?
Every post maps to a system Finzarc put into production, with the outcome attached, not roundups or predictions. It's the same principle as the rest of the site: working software over promises of future.