Everyone else raced toward the new. She turned around and walked straight into the oldest code in the building.
Pick the least glamorous problem in software. The one with the consultants, the spreadsheets, the decades of patches nobody dares touch. Now build the smartest AI in the world to fix it. That is the bet Emma Qian made when she left Google DeepMind.
Qian is the co-founder and CEO of Nova Intelligence, an agentic AI platform built for one specific, enormous job: analyzing, modernizing, and generating the custom code that companies have piled on top of SAP for decades. SAP processes an estimated 77% of the world's business transactions. Almost none of that runs on clean, documented code. It runs on layers - additions written by people who have since retired, for reasons nobody wrote down.
Most founders coming out of frontier AI labs chase the blank page: new apps, new models, greenfield everything. Qian looked at the immutable, impossible-to-rebuild plumbing of global commerce and saw, in her words, "the ultimate opportunity." The harder it is to replace, the more valuable it is to fix.
The resume reads like a straight line toward AI stardom. Caltech for computer science, with an exchange stint at Cambridge. Software engineering at Facebook AI, now Meta AI. Then Google DeepMind, the research lab that turned protein folding and game-playing into household news. Along the way she won first place at Outside Hacks VI and took first worldwide at the Data Open.
She had also already done the founder thing once - co-founding Multi, a foundation model and orchestration platform that raised $10 million. So the next move could have been anything shiny.
Instead she fixated on a number that sounds like an accounting error: for SAP, companies spend roughly seven dollars on consultants for every one dollar they spend on the software itself. The economics are broken, and they are broken at planetary scale. AI can write a brand-new app in an afternoon. It cannot casually rebuild the system that runs a Fortune 25 supply chain. That gap - between what AI loves to do and what the world actually needs done - is exactly where Qian planted her flag.
She did not try to replace SAP. Plenty of giants have tried; the graveyard is full. The insight was quieter and sharper: make the existing system work radically better, with AI agents purpose-built for it.
Then she assembled a team that could only have been built on purpose. Sam Yang, a repeat operator, as co-founder. And Professor Alexander Zeier - co-inventor of SAP HANA and former CTO of Accenture's SAP Business Group - as the third co-founder. To round out the credibility, Justin Kershaw, the former CIO of Cargill, a Fortune 25 company, signed on as chief customer officer.
It is a roster that says: we know exactly how deep this rabbit hole goes, and we brought the people who dug part of it.
"Emma is effectively the smartest person I know. The SAP problem felt genuinely hard."
Nova's strength is bridging two worlds: the rapidly evolving field of AI and the mission-critical reality of enterprise systems.EMMA QIAN, CO-FOUNDER & CEO
Nova's agents map decades of custom SAP code - the patches, exceptions, and forgotten logic that no current employee fully understands.
It refactors and untangles custom code so enterprises can move toward a "clean core" and SAP's S/4HANA without starting from zero.
From design through development to testing, the platform generates and validates code across the full migration life cycle.
The timing is not an accident. SAP's customers face a mandatory migration off legacy systems to S/4HANA, a deadline extended to 2030. Analysts size the broader wave at roughly $89 billion. Nova is built precisely for that crossing.
Festo, a customer that has run SAP since the 1970s, estimates it can now work roughly five times faster - complex programs that took months now finishing in days. Nova's case studies point to dramatic cuts in the manual grind.
Figures reported by Nova Intelligence and its customers.
She left frontier AI research at DeepMind and Meta to work on legacy enterprise software - the exact opposite of where most AI founders run.
Her co-founder Prof. Alexander Zeier helped invent SAP HANA, the database at the center of the system Nova is built to fix.
Nova's chief customer officer is Justin Kershaw, former CIO of Fortune 25 company Cargill.
She studied computer science at Caltech and spent an exchange term at the University of Cambridge.
First worldwide at the Data Open; first place at Outside Hacks VI - a habit of winning the competitions.
One early customer, Festo, has been running SAP since the 1970s - older than most of the people now modernizing it.
The future of AI, in Qian's telling, is not only about generating the new. It is about untangling the old - the systems too important to break and too tangled to leave alone. If she is right, the most interesting problem in artificial intelligence was hiding inside the most boring software on Earth, and Nova Intelligence intends to be the default way enterprises cross the bridge to 2030.
Sources: Nova Intelligence, Fortune, Founders YSK, LinkedIn. Facts drawn from public reporting and the company's own materials.