How I Work
One method, from first conversation to go-live
Roadmaps, enterprise deals and prototypes look like three different jobs. They fail in the same way: someone approves a summary instead of the system. So I run all three on one loop.
The loop
Study the system as it runs
Map how orders, stock, catalogue and money actually move today, including the manual workarounds and the worst day of the year. Averages hide the peak, and the peak is where software fails.
Name the fit, the configuration and the gap
Mark every requirement as fits, configures or does not fit. The stated misfit is the one claim a buyer always believes, and it earns trust faster than any demo.
Prototype on real data
Build the rough version on the customer's own products, volumes and exceptions before anyone commits. It answers questions a deck cannot, while they are still cheap.
Ship with the team, not around it
Hand the prototype, the evidence and the decision back to the team, and ship through the normal review, testing and ownership.
Reconcile sold scope with delivered scope
Six months in, compare what was sold with what people use every day. Every gap becomes the next item on the roadmap.
Building with AI
Agents do the volume. I own the judgement.
AI agents are part of how I work every day. They make producing work cheap. They do not make trusting it cheap, so I split the work deliberately.
| Work | What the agent does | What I own |
|---|---|---|
| Research | Gathers and summarises sources | Which sources deserve trust |
| Requirements | Drafts stories and edge cases | Scope, sequence and priority |
| Prototypes | Writes and refactors the code | The test that proves it works |
| Analysis | Runs the queries, builds the tables | Whether the answer makes sense |
| Writing | Drafts structure and prose | The argument and the final edit |
Leadership
Scale the judgement, not the headcount
I lead more than 100 people, directly and through the leaders who report to me, across the teams that design, build, sell and deliver the platform. An organisation that size cannot wait for one person to approve every decision, so I design it to decide well on its own.
One outcome per team
Every team knows the single result it owns and how it will be measured. Activity is not progress until that result moves.
Authority with ownership
The person closest to the problem makes the call. My job is to set the boundary and the standard, not to sit inside every decision.
Evidence before status
Reviews start from the raw data, the customer record and the working build, not from the slide that summarises them.
Leaders who build leaders
I coach leads until the method no longer depends on me, and I judge them by the leaders they develop in turn.
Escalations stop with me
When something breaks for a customer, I take it personally, so the team can keep building while it is fixed.
Agents beside every team
AI agents take on the volume of research, drafting and testing, with one shared standard for checking what they produce.
Questions I am often asked
Do you still write code?
Yes. I am a programmer, and I build products myself, including a payment reconciliation engine, a jewellery ERP and returns apps for Shopify and Adobe Commerce, working with AI agents. Anything that goes to production goes through the team's normal review. See the products I have built.
How deep is your AI expertise?
Deep enough to build with it, govern it and teach it. I understand AI from the model up: models and their failure modes, context, retrieval, tools, agents, evaluation, cost and governance. I built my most recent products myself with AI agents, run several coding agents in parallel every day, and wrote the reusable skills, quality gates and repository rules that let agents work to my standard. Inside the products I lead, AI enriches catalogues, matches payments and powers virtual try-on. I work at every stage of the ten-stage model I wrote about, up to shaping how whole teams work with AI. See where I work on the ten stages.
With more than 2,500 capabilities shipped, is this a feature factory?
No, and the number is the proof rather than the point. Every one of those capabilities began as a named problem: an order a brand could not route, a settlement a seller could not reconcile, a return a shopper could not complete, a store that could not see online stock. A request becomes work only when we can say who has the problem, what it costs them and how we will know it is gone. Anything that cannot pass that test waits, however loudly it is asked for. Across eleven years and more than 1,000 brands, that discipline is what keeps every release tied to an outcome a customer can feel.
How do you lead a team of more than 100 people?
Through leaders, not around them. I lead directly and through the leads who report to me. Each team owns one outcome, each lead holds the authority to reach it, and every review starts from the raw evidence. I hire for judgement, coach until the method no longer depends on me, and take the hardest escalations myself.
Which problems do you enjoy most?
The ones where operations, systems and money meet, anywhere from manufacturing and procurement to the delivered parcel: order flow, stock accuracy, integrations, B2B ordering, payments and reconciliation, and the business case behind each of them.
How do you run enterprise pre-sales?
As the first product review. A system study, a fit table in every proposal, prototypes on the buyer's data, and the same commitments before and after signature.
Where have you worked with enterprises?
India, the Middle East, South East Asia, Hong Kong, South Africa, the United States and the United Kingdom, with more than 1,000 brands, retailers, marketplaces, logistics companies and distribution partners, across more than 500 channels and integrations.
Which brands have you worked with?
Brands such as Nykaa, Metro Cash & Carry, JioMart, Titan, Bata, Pepe Jeans London, Aldo, ASICS, The Body Shop, Kiehl's, Lulu Hypermarket, Amazon Business and Swiggy Instamart, and platforms from Amazon, Walmart and eBay to Shopify, BigCommerce and Adobe Commerce. The full picture is on the clients and partners page.
What do you write about?
AI, product and commerce systems, in The Rigorous Builder, a journal I edit.
Work with me
Let's build what comes next.
I work with operators, product leaders and founders on commerce platforms, marketplaces and building with AI. A short conversation is the fastest way to find out whether I can help.