Glen Ellyn, IL, USA, October 12th, 2026, FinanceWire
Uday Surendra Yandamuri
Paper presented at the Third International Conference on Innovations in Cybersecurity and Data Science proposes a layered, human-centered framework for adaptive enterprise automation and governance
Uday Surendra Yandamuri, a technology and operations analyst with a background spanning hospitality technology and agricultural technology, has co-authored research proposing a new approach to enterprise automation: information systems that adapt to changing conditions rather than staying fixed once implemented. The paper, titled “Adaptive Intelligence Networks for Human-Centered Enterprise Automation and Governance,” was presented at the Third International Conference on Innovations in Cybersecurity and Data Science and published through IEEE Xplore.
Most business processes are designed to remain stable once put in place, even as organizations change through mergers, expansions, and shifting cost pressures. That stability offers speed and predictability, but it leaves little room for systems to respond when conditions shift unexpectedly, whether from a global disruption or a sudden spike in demand. The authors argue that enterprise automation needs a layer of adaptability that most current information systems were never built to support.
Their response is the concept of Adaptive Intelligence Networks: layered systems in which specialized components are trained to optimize decisions across different timeframes, locations, and information contexts. At the top of the structure sit Master Adaptive Systems, which set locally optimal governance decisions for a set of peer Adaptive Authority Systems. Those authority systems assign rules to the individual agents operating within their scope, while also coordinating with other Master Adaptive Systems to manage resilience, information flow, and security across the broader network.
The research grounds adaptability in specific mathematical formulations for system utilization, latency reduction, and resource efficiency. It also introduces the Functional Adaptation Intelligence metric, which combines explainability, traceability, and accountability into a single measure of system trustworthiness. The architecture spans four layers, sensing, contextualization, dynamic innovation, and enterprise-process support, each evaluated separately for latency and resource use.
Key findings reported in the paper include:
- Latency reductions of 28 percent at the sensing layer to 31 percent at the enterprise-process layer, compared with static baseline systems.
- Over an 18-month evaluation, the adaptive network’s F1-score climbed from 72.5 percent to 86.9 percent, while a comparable static system plateaued near 74 percent.
- Average CPU usage cut by roughly 27 percent and memory usage by about 33 percent versus standard distributed architectures, with CPU reductions as high as 40 percent at edge nodes through more efficient task scheduling.
- Higher throughput than static distributed systems as node counts grew, which the authors attribute to the network’s ability to reorganize its function distribution rather than simply adding capacity.
- Lower error rates and faster recovery across five operational scenarios (normal conditions, peak load, fault events, network degradation, and resource scarcity), including a 57 percent reduction in errors and a 66 percent improvement in recovery time under peak load.
- Performance on par with cloud-only scaling while reducing operational costs by roughly 22 percent.
Despite the technical depth of the architecture, the paper returns repeatedly to human-centered design as a core requirement. The authors argue that adaptive systems deployed in governance settings must remain transparent enough for operators and managers to understand how decisions are reached, cautioning that interaction through a user interface alone may not be sufficient to build the trust these systems require. In some cases, they suggest, meaningful oversight may call for direct sessions in which operators work through how a system’s internal models relate to the decisions it makes.
Yandamuri’s research was also recognized this year with a Best Paper Award at the International Conference on Computing and Communication Networks (ICCCNet-2026), for a related paper titled “Real-Time Data Analytics in Smart Networks Using LLMs and Cloud-Based Computing Frameworks.” The recognition points to a consistent focus across his work: combining real-time data processing with intelligent, adaptive infrastructure to support faster and more resilient decision-making.
The paper closes by acknowledging that adaptive intelligence networks are not yet widely deployed, in part because established design principles for building them remain limited. Yandamuri and his co-authors position their framework as a step toward closing that gap, offering a structured, measurable approach to adaptability that spans everything from individual user devices to enterprise-wide delivery chains, while keeping explainability and accountability built into the system rather than added on afterward.
The full paper is available through IEEE Xplore.
About Uday Surendra Yandamuri
Uday Surendra Yandamuri is a technology and operations analyst with a background spanning hospitality technology and agricultural technology. He is a co-author of “Adaptive Intelligence Networks for Human-Centered Enterprise Automation and Governance,” presented at the Third International Conference on Innovations in Cybersecurity and Data Science, and a recipient of a Best Paper Award at ICCCNet-2026.