Pujan Motiwala
Founder and Principal, Ahrom Labs. I map how a business actually runs, then build the system it runs on, and I'm the person who answers when you write in.
The map comes first
I learned to draw how work moves before I wrote software for anyone. During my master's in Information Systems at Stevens Institute of Technology, where I studied business process management, I mapped the workflows of a defence research centre: how each request moved, where it waited and what it should look like instead. That habit is the first phase of every Ahrom Labs engagement. We draw your business as it is before anything is built, because a system built on the wrong picture only makes the workarounds faster.
Nine years inside other people's numbers
My work has been where the data meets the money. I analysed insurance claims at Bajaj Allianz and customer data at Liberty Mutual. At USAA I built predictive maintenance and asset optimisation models, at Arkose Labs fraud and bot detection, and at ABC Fitness churn and retention models across their clubs. I also taught Python for data analysis in Columbia University's Applied Analytics programme.
Every one of those projects taught the same lesson: a model is only as good as the records under it. Most Indian manufacturers and traders don't have an AI problem yet. They have a records problem: the truth is split across Tally, spreadsheets and a few people's memory. So that's where we start.
What I build at Ahrom Labs
Since January 2025 Ahrom Labs has built and run the operations and accounting systems for an electronics component trader, an interior design studio and its furniture workshop, and a boiler manufacturer: sales to dispatch, stock, statutory documents, payroll, TallyPrime posting, and AI that reads invoices and answers questions from the books, with a person confirming anything that matters. The systems are in the work; how they were built, and why, is in the engineering notes and patterns, every one written by me.
Where your system goes next
Once your records live in one structure, the work I did for banks, insurers and security teams applies to your business too. These are the next steps on a system you already run, and none of them needs a new one.
Machines and jobs that are about to slip
Your job cards, breakdowns and dispatch dates already hold the pattern. A model trained on them can flag the machine due for trouble or the order likely to run late, while there is still time to act.
Cash and stock, before the month closes
Receivables, payables and stock movement in one ledger can be projected forward: what comes in, what goes out, and which item runs short first.
The entry that doesn't fit
The methods banks and security teams use to catch fraud work on a factory's books too: a rate far off the purchase order, a payment out of pattern, stock that moved without a document. One version already runs: every vendor and freight charge is checked against its agreed rate.
Customers going quiet, or due to reorder
Order history shows who usually buys when. The system can tell sales who has gone silent and who is due, before a competitor calls them.
Questions answered from your own records
Ask in plain words, in English or Hinglish, and get the number from your own data with its source. Invoices, certificates and drawings are read for you and checked by a person before they post.
The rule doesn't change as the AI grows. The system suggests, a person confirms, and nothing posts to your books on a guess.
Elsewhere
- Education
- Master's in Information Systems, Stevens Institute of Technology, USA (2016). Bachelor's in Information Technology, Gujarat Technological University (2014).
- Also founded
- ClickCatalyst, a separate company that builds Google Ads audit software.
- Profiles
- LinkedIn · GitHub · Medium · pujan@ahromlabs.com
Talk to us about your systems.
Tell us what's held together with workarounds right now. We reply to every message ourselves.