Product manager. Systems builder. AI realist.
Thirteen-plus years inside operations, finance and logistics. Almost nine of them shipping the internal products that run them. Process first, then product, then AI, in that order. Based in Gurugram, India. Working with teams anywhere.
The short version
I build internal products, automate the processes around them, and put AI in only where the data can carry it. I have done this inside a startup, inside a large logistics company, and now as a consultant, and the pattern has never changed: sit with the people doing the work, find where it breaks, put the process into a system, ship it, then watch who actually uses it.
Instaspaces was the proving ground. I joined a small meeting-room startup running accounts and operations when GST changed the market, and led the virtual-office product from research to launch. Then the booking database, the approvals, the vendor dashboards, the CRM, all rebuilt as proper web apps on Apps Script. Six years of turning every function into a system.
CJ Darcl Logistics hired me to do it at scale: voice agents that called vendors, an AI copilot for a forty-person tracking team, a sales CRM on tools the company already paid for. That is also where I learned the uncomfortable part: a proof of concept can hit every number and still get parked if nobody owns it.
The habit underneath all of it comes from my first four years, in audit and financial reporting: if a number cannot be traced, it is not a number. Every board I build is designed so nobody has to take it on trust.
The journey
Scroll, and the line draws itself. Or pick a stop.
2013
Where I learned that a number you cannot trace is not a number.
Staff accountant at Gopal Rao & Co. and Jain S Kumar & Associates. Financial reporting, accounting, tax filings and audits for domestic and international clients. CA Intermediate in 2013, B.Com (Hons) from Delhi University in 2016.
2017
Meeting rooms by the hour. Then GST changed the market.
Joined a startup renting meeting and training rooms, running accounts and operations. GST had just launched and businesses needed registered addresses in every state. We saw the gap and started building for it.
2018
Market research, product fit, demand and supply.
Defined the virtual-office product end to end: registered addresses for business registrations, order to service delivery, and the booking database behind it, built on Google Workspace. The first system of many: approvals, vendor dashboards, finance reporting.
2020
Offices shut. Registered addresses did not.
When lockdown made co-working spaces and office buildings non-operational, demand for virtual offices moved online overnight. Two years of product groundwork meant we were the ones ready for it.
2022
Designing myself out of the loop.
Rebuilt the internal systems as Apps Script web apps on a Sheets backend: the booking dashboard about five times more efficient, the sales funnel turned into a CRM, expense approvals over WhatsApp, a company-wide KRA/KPI framework, and a billing process that ran without me except for escalations. Trained with CEOITBOX (2021) and BCI (2022) along the way; CEOITBOX's Excellence in Implementation award followed in March 2024.
2024
Business Excellence. AI agents at scale.
Joined CJ Darcl Logistics to lead product, automation and AI initiatives. Voice AI for vehicle hiring, an AI tracking copilot, a sales CRM with geo-tracking, and AI knowledge-sharing across teams. Worked directly with the President's office.
2025
Every number hit. Nobody owned it.
The voice hiring agent lifted vendor pick-up from one in five to three in four. After I left in November 2025 it had no owner and was parked. It is the reason I now insist on ownership and gates before any AI spend.
Dec 2025
Process, product, then AI, for anyone who needs it.
Started consulting. The pitch is simple: I find where your operation breaks, put the process into a system, ship the product, and only then bring AI in.
2026
From 30 seconds a save to under one.
Built an own-fleet operations platform for a logistics business unit, first on Apps Script, then rebuilt on Firebase when the team said they would only adopt it under two seconds. Whole fleet on it within a week.
2026
Two more products, one with AI inside.
Chakshu captures daily indents and placements for a large consumer-goods client; MailSutra turns the chairman's emailed ideas into a tracked initiative register with AI-written summaries. Both live.
Selected work
14 books on the shelf, coloured by the kind of work. Pull one out to read the brief and watch the demo, then open the full case study if it earns it.
Nothing on this shelf for that filter. Try another.
Pick a book to open it.
Why I build for mental health
Ritmann is the brand my wife and I run together. We talk about relationships, couple mental health, travel and the ordinary business of living with each other. Steady is its product.
Steady is a free, private mood journal for adults living with ongoing emotional-health conditions, and later for the partners who love them. Most trackers interpret, nag and harvest. Steady does the opposite. Logging a feeling takes ten seconds. It is observational, never interpretive. Skipping a day is still participating. Everything stays on your phone unless you choose to back it up to your own Drive. No streaks, no red marks, no ads, no analytics.
Toolkit
Hover or tap a tool.
How I work
FleetSetu, step by step. It is how I approach most problems. Watch the thing being built change shape on the left.
The brief said "optimise the fleet report". The trackers said the report was the last link in a chain of retyping.
784.6 man-hours a month went into manual reporting. One person updated 400 vehicle statuses a day at 30 to 40 seconds each.
Every change was discussed first and signed off in writing. One chunk at a time. Nothing silently regressed.
One status model instead of three vocabularies. Detention derived from timestamps, so nobody can declare it. GPS tags position; a human sets status.
The team said under two seconds or nothing. Rebuilt on Firebase: under one second. Whole fleet onboarded in a week.
Trip intelligence, an AI report analyst and voice alerts sit behind gates on the roadmap. The data layer earns them.
Three rules I work by
Discuss, sign off, then build.
Fix data before buying AI.
Gates before spend.
Automation, to me, means a process runs on its own with minimal constant monitoring, because the system does the monitoring. I learned the "business autopilot" idea from Rahul Jain's coaching and have been applying it since 2022.
Before and after
Every messy process is a system waiting to be built.
After hours
Born in Mumbai, Delhi since 2006, Gurugram since 2021.
Couple content, mental health, travel and lifestyle, with my wife. @rit.mann
Guitar, piano, and I sing. Badly in the car, better with a mic.
Valorant and League of Legends. I post gaming content as broitsmanhar.
I teach myself AI using AI. Every week something I believed last week is wrong.
I cannot stop simplifying other people's daily tasks. MailSutra started as one of those favours.
The board
Say hi, tell me what keeps breaking in your business, or what you are building. It goes up on the board, and I read every one.
Pinned. It's on the board.
Contact
Open to consulting with businesses that want their operations in a system before they buy AI, and to senior full-time product or operations roles with companies I can learn from. I reply fast.