Proof of work · AI advisory
Two days inside a dairy company, rethinking it with AI.
In August 2026 I got to work with Dodla Dairy — one of India's largest listed dairy companies — on an engagement I'd been dreaming about: not “let's automate something,” but let's sit with the whole leadership team and rethink what this company is, now that AI exists. Here's what we did, and what it taught me.
§ 01 · The shape of it
What we actually did
The engagement had a simple spine: understand the company, build a thinking tool, then get the whole leadership into a room and work.
Before
The recce
Discovery calls across departments, reading everything public, and mapping how the company actually runs — who owns which decision, and which document each person opens every morning. That last question tells you where the real data lives, and where it doesn't.
The demo dashboard
I built a high-fidelity, fully synthetic demo of what an AI-era operating layer could look like for them — group view down to departments, plants, procurement, sales, logistics — with AI agents wired into every page. Deliberately more extensive than a workshop needs, because the value of an AI-era dashboard isn't in any one page. It's in the connections between them.
Day one
Leadership session — the frame
A morning on what AI actually is (a reasoning ability, not a software category), then a walkthrough of the demo end-to-end with the senior team reacting to specific views and specific agent actions. Concerns collected, source-of-truth documents on the table.
Day two
Department deep-dives
Three-hour working sessions with individual departments — problem-finding first, then building. Each session ends with a named output: a list the team leaves with, not a deck they file away. The demo dashboard becomes the baseline to improve, edit, and argue with.
The handover
Everything they need to carry it forward without me: the session materials, the write-ups, the build prompts to replicate the demo, and honest notes on what a production version would take. A good engagement should end with the client more capable, not more dependent.
§ 02 · The demo
A dashboard as a thinking tool
The demo ran on one adoption idea: every senior leader has one daily headache the dashboard solves — that solution is why they actually open it. Underneath, every page feeds a shared intelligence layer, and AI agents act inside that context: drafting interventions, creating tasks, flagging anomalies that cross silos. The move I wanted the team to feel is the dashboard going from report to operator.
And an honest disclaimer I kept repeating, because it's the point: the demo doesn't touch their real systems and every number in it is synthesised — accurate in shape, not in reality. It's a thinking tool. A way to compress a year of “what if we had X?” into something a leadership team can click through together in a workshop room, and react to concretely.
Today · exception-first
The morning brief
Recreated in CSS, numbers invented — the shape of the thing, which is exactly what a thinking tool should be.
§ 03 · The night before
The dashboard was never the point
Here's the part I'll remember longest. The company is so rich and multi-faceted that I spent weeks happily engrossed in the game of fitting it all into a dashboard — and the team played along with me. The night before the first session I had to physically snap out of it and say: no. The dashboard is one output. It is not the purpose.
The purpose was to rethink the company in the context of AI — to ask what it means to run a three-decade-old business when reasoning itself has become cheap. The framing that anchored the whole workshop:
“AI doesn't augment your company. It reveals what your company actually is.”
For this company, what it reveals is lovely: their edge was never just the product — it was understanding one person, the farmer, better than anyone else, and solving for what they needed. The AI-era question is simply that same edge, at scale: the same depth of understanding, extended to every farmer, every customer, every exec on the ground — heard in their own words, in their own language. What we don't measure, we can't improve. So each working session ended by naming what to start measuring, and who's slipping away while we chase the new.
§ 04 · Field notes
What I learned about rooms
The technical build matters, but engagements are won or lost in the room. These are the notes I'm carrying into every future workshop.
The frame beats the artifact
Whatever you build, it will try to become the purpose. Guard the frame ferociously — the artifact is just the excuse to get everyone thinking in the same direction.
Here: “rethink the company with AI” survived; “admire the dashboard” didn't.
Champions compound
One internal champion gets you in the door. But when the seniormost people in the room lean in, something shifts — suddenly people are being called into the room mid-session.
Here: department heads were offered a 1-hour version of the deep-dives. They chose the full 3 hours. That's the real yes.
Every room has five people
The next-gen champion. The founder-leader with the most knowledge and the most gravity. The reluctant one whose conversion signals success to everyone. The quietly confident one. And the information powerhouse who stays silent unless explicitly invited — often because of language.
Here: learning to fuel the session with all five, instead of managing around them.
Say the language thing up front
An announcement at the start — share in whichever language you're comfortable in — costs ten seconds and can unlock the most knowledgeable person in the room.
Here: my clearest improvement for next time. AI handles vernacular beautifully; the room should too.
The best sign: they ask your question
When a leader turns to a colleague and organically asks the exact question you were building toward, the session has become theirs. That's the goal — a conversation that no longer needs you.
Here: it happened before I even reached the brainstorm slide. Best moment of day one.
Cross-functional is a choice, not a default
With a large company the danger is fragmentation — you could scope tighter and go deeper. But a company that already has data and views everywhere may genuinely need the connective, cross-functional layer first. Match the scope to the company's stage, not to your comfort.
Here: they had the dashboards. What was missing was the intelligence between them.
- 🐄 A cow's muzzle print is unique — nature shipped biometric ID first.
- 🔬 Milk is tested on 40 quality parameters before it reaches you.
- ❄️ It's chilled to under 5°C minutes from the farmer, via heat exchanger.
- 🏭 They implemented SAP unusually early — an edge that explains their data culture to this day.
- 🥛 A tour of the milk process here is called a “Paal Yatra” — and yes, it's as delightful as it sounds.
- 📱 Field executives are a high-touch army — there are more milk-field-tracking apps than you'd believe.
§ 05 · What I'm selling, apparently
It was never just AI
A framing from the company's leadership that I've adopted wholesale: they didn't bring me in for AI knowledge alone. They brought in the whole stack — years inside a family agri-business, building and running Hoovu, Stanford, the operator's instinct for what a morning actually looks like in a business that moves physical goods. AI is one piece of that puzzle — the newest one.
The honest priority order of what an engagement like this delivers:
§ 06 · With thanks
Grateful, mostly
Dodla Dairy is among the top listed companies in the country by revenue, thirty years old, built on a genuinely beautiful founding insight about trust and the farmer. A company like that has every excuse to move slowly and every right to be sceptical of a consultant with a laptop full of ideas. Instead, they were the warmest, fastest room I've worked: leadership who sat through every session, department heads who chose the longer version, and a team that argued with the demo in exactly the way you hope people will argue with a thinking tool.
I left with a camera roll of milk-glass toasts, a notebook of things I'd do better, and the strong suspicion that this — sitting with real operators, rethinking real companies — is the most fun version of working in AI right now.
“The dashboard is an output. The reframe is the product.”
This is the shape of the ₹5,00,000 session tier — a customised day with a demo dashboard built for the company beforehand.