Sivasankararao Kavuri.
Eighteen research articles in seven years have charted how companies move and govern their SAP data. His latest work asks how artificial intelligence can take on the most stubborn manual tasks, without taking people out of the loop.
For most of the past decade, the hardest work in enterprise data has been done by hand. Analysts map thousands of legacy fields to new ones. Testers sort loaded records into valid and invalid. Business users wait for specialists to answer basic questions about customer and material records. Sivasankararao Kavuri's latest research asks how much of that work artificial intelligence can now share.
Kavuri, who studies how large organizations migrate and manage the business data inside their SAP systems, published his eighteenth research article in August 2026. Over the past year, his work has focused squarely on applying AI to the slowest and most error-prone steps in that process.
The timing is deliberate. Companies are racing to complete SAP S/4HANA programs before the 2027 end of mainstream maintenance for ECC. They are also being asked to show returns on AI. Both goals depend on the same thing: business data that is accurate, consistent and trusted.
Three questions for the AI era
Kavuri's recent studies can be read as answers to three practical questions.
The first is whether generative AI can help with data mapping, the painstaking task of matching every field in a legacy system to its place in S/4HANA. His research evaluates AI-assisted approaches that suggest likely matches and transformation rules for analysts to review, aiming to cut weeks from one of the longest phases of any migration.
The second is whether machine learning can speed up the checks that follow each data load. Every migration cycle ends with teams classifying records, explaining failures and routing them for correction. His work explores how that classification can be automated, so problems surface sooner and repeated issues are caught as patterns rather than one record at a time.
The third, and newest, is whether business users can work with master data directly, through conversation. His August 2026 research looks at how chat-style interfaces can let staff ask about records, track requests and start approved changes in plain language, while the controls that protect the data stay firmly in place.
These studies build on his earlier work using machine learning to help design data cleansing rules, which first brought AI into his research in early 2025.
Keeping people in charge
What sets Kavuri's AI research apart is its restraint. Each study treats AI as an assistant to experts, not a replacement for them. A human analyst approves the suggested mappings. A data steward confirms the automated classifications. A governed workflow sits behind every change requested through a chat window.
That approach reflects lessons from the rest of his research. Before turning to AI, Kavuri had already published work on access governance during migrations, on keeping test and production systems synchronized, and on approval and stewardship models for centralized master data. His AI studies extend those safeguards rather than bypassing them.
For industries where data errors carry regulatory or financial risk, that balance matters. It offers a path to adopting AI in data management that compliance teams can support.
Seven years of groundwork
The AI papers rest on a broad foundation. Since his first article in 2019, Kavuri has covered almost every stage of an enterprise data program.
His early research dealt with speed and quality: starting S/4HANA migrations faster with pre-built content, and tracking data defects through every test cycle against clear service levels. He then turned to connection, studying how on-premises systems can feed cloud data lakes in near real time.
From there his work moved into the detailed governance that experienced teams know decides success: value mappings that must match across environments, cross-reference tables that link old keys to new ones, and the partner relationships that define complex customer and vendor hierarchies. Later studies reached into planning data that crosses ERP boundaries, serialized warehouse stock, transportation and route data, and batch exchange among multiple systems.
Seen together, the research follows a clear arc: first moving data, then connecting it, then governing it, and now using AI to manage it more intelligently.
Why his work matters
Kavuri's main contribution to the field is the way he turns hard-won project experience into methods others can reuse. Much of what separates a smooth migration from a troubled one has never been written down. It lives with the specialists who have learned it the hard way. His papers put that knowledge on the record.
The impact reaches several groups at once. Migration teams gain tested approaches for the steps that most often cause delays. Data governance leaders gain models that extend well beyond go-live. And organizations planning for AI gain research that looks at both sides of the equation: how to prepare reliable data for AI, and how AI can help create and maintain that data.
As AI tools move into everyday enterprise software, that second question is becoming more pressing. The organizations that succeed are likely to be those that pair the technology with clear accountability. Kavuri's research has been making that case consistently.
What comes next
With eighteen articles published and his newest work focused on conversational AI, Kavuri's research is moving closer to the people who use enterprise data every day, not only those who manage it behind the scenes. If his record is any guide, the next stage will keep the same focus: practical methods, built from real problems, that help organizations trust their data.
