Irfan Khan has a fondness for metaphors that make spreadsheets feel almost athletic. Enterprise data should dance. Supply chains need oxygen. Intelligence deserves an assembly line. After 30 years spent around business systems, he speaks about the invisible plumbing of companies as if it might step out for an encore. The language is lively; the conclusion beneath it is stern. Artificial intelligence becomes useful only when somebody does the unglamorous work of connecting it to reliable information, everyday workflows and a person willing to own the result.
That conviction now shapes CLOUDSUFI, the San Jose enterprise data and AI company Khan co-founded in 2019 with Zeba Vakil. Khan is its president, CEO and board chairman. The firm began with a compact purpose: reduce the distance between human intuition and data-backed decisions. In 2026, that premise has acquired new machinery. Khan talks less about isolated proofs of concept and more about AI factories, data factories and control towers - repeatable systems that help a company move from an amusing demonstration to production work.
The distinction is easy to miss in a season of theatrical software. A demo needs a clean prompt and forgiving lighting. An enterprise system must survive permissions, old databases, surprise formats, audits, quarterly budgets and the colleague who has sensibly kept a critical process in a spreadsheet since 2009. Khan’s argument is that the model is only one station on the line. The actual product includes the data feeding it, the connectors reaching other systems, the rules limiting its reach and the measurement proving that it earned its chair.
The supply chain apprenticeship
Khan arrived at the AI boom by the long road through supply chains. He joined Bristlecone around 2002 and spent 12 years moving through the company before its board appointed him president and CEO in September 2014. His previous post was senior vice president of global field operations, the sort of title that suggests fewer stage lights and more airports. By then, his remit had crossed the Americas, Europe, the Middle East and Asia. The work put him close to the awkward border where a strategy slide becomes somebody else’s Tuesday.
In a 2015 interview, he described analytics as an evolution. It had moved from descriptive reports to prescriptive and predictive systems, then toward what the industry called cognitive analytics. Bristlecone’s earlier work had relied on structured data and even Excel sheets. The newer systems absorbed messier signals and attempted to anticipate demand or competitive behavior. Khan summarized the ambition with typical directness: it was like predicting the future today.
The prediction did not quite erase the operational world. It made that world hungrier for data. A forecast could be clever and still be useless if the information arrived late, lived in incompatible systems or could not reach the decision-maker. Supply chains are merciless teachers of this point. Trucks, parts and purchase orders cannot be persuaded by a keynote. They expose every broken handoff.
Joins Bristlecone and begins a 12-year rise through global operations.
Appointed Bristlecone president and CEO.
Co-founds CLOUDSUFI around enterprise data and decisions.
Pushes AI factories as the bridge from experiments to industrialized delivery.
Khan’s education fits the double view: a Bachelor of Engineering from the College of Engineering and Technology in India, then an MBA from S.P. Jain Institute of Management and Research. His career also includes leadership work at Microsoft and Hughes Software. Engineering helps him see the system; operating experience keeps asking who will use it and why.
“Every enterprise needs assembly lines for intelligence.”Irfan Khan on industrialized AI delivery
Factories, not magic tricks
At CLOUDSUFI, the assembly-line idea is literal enough to be useful. Khan describes modular components that can be combined instead of rebuilt for each customer. A deployment may include standardized generative AI patterns, prompt-pipeline governance, model-lifecycle management, use-case-level return tracking and enterprise guardrails set at the beginning. Connectors become their own factory because serious AI needs to communicate with dozens, sometimes hundreds, of existing applications.
The word factory is deliberately unfashionable. It replaces the romance of a singular breakthrough with repetition, quality control and accountable output. Khan says CLOUDSUFI is moving from random customer projects toward industrialized services. Deals can run for one or two years. Milestones may change as the work reveals what is possible, but the reusable foundation remains. The model resembles Lego, another Khan analogy: standardized pieces that save teams from manufacturing the same brick each morning.
His sharpest current position concerns the raw material. Khan argues that companies should train enterprise systems on controlled, credible sources rather than scoop indiscriminately from the open internet. He has called internet data “poison” in this context. The phrasing is blunt, but the concern is familiar to anyone responsible for a corporate decision: provenance matters. A system cannot become trustworthy by ingesting information whose quality and permissions nobody can explain.
This emphasis turns AI strategy into an exercise in boundaries. What information may an agent access? Which tools may it use? What can it execute alone? Where must a person approve? When should its access expire? Khan’s recent public writing places people at the top of that chain. Humans set direction. AI actors execute at scale. The point is not to preserve a ceremonial human in the loop. It is to make authority visible before speed makes mistakes expensive.
A company with verbs
CLOUDSUFI’s own vocabulary is unusually kinetic. Its mission speaks of distilling noise, automating data supply chains and making enterprise data dance. Khan calls data the new oxygen. The phrases could float away if the business beneath them were vague. Instead, they keep pointing to the same operating sequence: collect what is trustworthy, connect what is scattered and return a decision while it can still matter.
Khan’s role in that system is part salesman, part translator and part field operator. He publishes on antifragility, supply chains and data strategy. He is an angel investor and adviser to enterprise software and services companies. He has belonged to the Forbes Technology Council since 2019, and his public profile also lists the Wall Street Journal CEO Council. Yet his most persuasive ideas are not predictions of machine omniscience. They are descriptions of organizational chores that technology can finally perform with more range.
There is a social ambition too. The company’s stated values are passion, integrity, empathy and boldness. Khan has written about the CLOUDSUFI Foundation as an effort to bring elders back into work with technology, pairing experience with new tools. His broader aspiration is a “human-powered by AI” enterprise. The phrase keeps the hierarchy clear: capability expands, responsibility does not evaporate.
“Data is the new Oxygen.”Irfan Khan’s shorthand for the enterprise dependency
The appetite behind the operator
For all the factory talk, one of Khan’s warmer public stories concerns food. He wrote about more than two decades of friendship and work with colleague Rajan Venkateswaran, a relationship maintained partly by finding good South Indian restaurants wherever business took them. They once missed a flight because a new place was too good to leave. Strategy apparently lost a close vote to sambhar.
The anecdote reveals a different kind of infrastructure. Khan’s career is threaded with long partnerships. Venkateswaran later became part of CLOUDSUFI’s leadership journey. Co-founder Vakil runs operations. Khan routinely credits employees, customers and partners when describing the company’s progress. Even his metaphors are communal: a factory is coordinated work, a dance requires timing and oxygen is shared.
That relationship-minded style may be useful as enterprise AI grows less like a software purchase and more like organizational renovation. The hard questions belong to several departments at once. Data teams know where the bodies are buried. Security teams know why the gates exist. Operators know which exceptions are actually the business. Executives decide which errors the company can afford. A model cannot convene that coalition simply by being impressive.
Khan’s latest expansion puts this philosophy closer to North American customers. CLOUDSUFI is building AI engineering capacity in Mexico, which he describes as a strategic part of shortening the distance between enterprise data, decisions and action. The geographic move mirrors the technical argument. Get the builders close to the operating context. Reduce the handoffs. Make accountability easier to locate.
When the system clocks in
Khan’s story spans three eras of enterprise technology: structured reports, predictive analytics and now AI actors capable of navigating several systems. Across them, his central question has barely moved. How does information become action without losing the judgment that gives it meaning?
His answer is becoming more architectural. Proprietary data supplies the ground truth. Reusable factories provide speed. Governance limits authority. People choose the destination. This is a less dazzling future than the autonomous corporation, but it has the advantage of resembling a place where one might safely work.
The next phase of AI will produce plenty of spectacle. Khan is betting that the durable companies will be built in the quieter interval afterward, when the audience leaves and the operator asks whether the demo can run again, with real data, under real rules, for a customer who expects it to be working tomorrow. At that point intelligence stops performing. It clocks in.