From General Coding Agents to SAP: Founding Nova AI

The Foundation: General Coding Agents

Nova Intelligence began with a question about the limits of AI for software engineering. Co-founder and CEO Emma Qian came at the problem from her background at Google DeepMind, where she applied AI to solve math problems, and was drawn toward a similar question in software. It was clear to her that AI for software engineering would only become more important over time.

Emma and co-founder Sam Yang — alongside what would become the early Nova Intelligence team — started a research project to push the limits of code generation. Their work produced strong results on coding benchmarks at the time and began attracting attention from industry.

Discovering the Real Frontier

That work surfaced something unexpected. The most critical code in the world — the systems the largest enterprises actually run on — sits in a completely different tech stack than what general agents are built for. There's a class of enterprise systems where this is especially true: platforms that have been at the center of the largest companies in the world for decades, with deep customizations layered in over time to fit each company's specific business processes. The work of maintaining and modernizing that custom code is a category of its own — and one that general-purpose AI tools weren't built for.

Emma and Sam noticed it first in the economics: every dollar spent on this kind of software drives multiple dollars in services. That ratio is unusual, and it pointed at something specific. This software is dramatically more expensive to maintain, more customized, and more resistant to automation than anything general-purpose AI was touching.

The more they pulled on the thread, the clearer the opportunity became. This wasn't a small corner of enterprise software — it was the operational core of many of the world's largest companies, and it was almost entirely underserved by the current wave of AI tooling.

Within that class, SAP stood out. It's the largest and most widely deployed — the system of record for 92% of the Global 2000, running finance, supply chain, manufacturing, and logistics for the companies that move the physical economy. It's also the one facing the most acute pressure right now.

Why SAP, Why Now

Several forces were converging at the same time:

• A mandatory, once-in-a-generation migration to S/4HANA by 2030, costing some companies over a billion dollars.
• AI capabilities crossing the threshold needed to handle complex, domain-specific code.
• A widening shortage of SAP talent as a generation of experts approaches retirement, just as demand for transformation work spikes

That convergence is what pulled Nova toward SAP specifically — and made clear that solving it would require a specialized approach. The team brought on Prof. Dr. Alexander Zeier, co-inventor of SAP HANA and former CTO of Accenture's SAP Business Group, as Chief Scientist and co-founder. Alexander had spent decades inside the largest SAP transformations in the world and had seen firsthand where the manual, consultant-heavy work piled up. By combining frontier AI research with decades of SAP domain expertise encoded into the platform, Nova set out to build what no general coding agent could: AI-native automation purpose-built for the most foundational systems in global business.