UBS CTO Elango Somasundaram argues enterprise advantage shifts from access to models toward embedding rich, governed context across architecture so agentic systems can reason, coordinate and deliver correct outcomes.
Context—proprietary processes, product rules, security and workflows—is the differentiator because foundation models and skills are widely available.
Agentic systems need guardrails and reasoning across every architectural layer: identity, connectivity, SDLC controls, compliance and security.
The developer role changes: engineers will encode business context, validate outcomes and design guardrails instead of only writing tickets and code.
# Hyderabad 2026
# Agentic AI vs earlier waves
Past waves included robotic process automation, cloud, copilots that predict code, and vibe coding that lets developers author via natural language. Agents executed narrowly defined tasks. Agentic AI adds an orchestration layer: it reasons about objectives, creates task graphs, assigns subtasks to specialized agents and follows through to completion.
# Why context matters
Access to powerful foundation models and prebuilt skills is becoming commodity. What differentiates outcomes is context: the enterprise-specific data, workflows, policies and constraints that tell an agent what counts as correct or safe in a given situation.
For regulated industries like banking, context includes approval processes, product rules, security policies and customer expectations. Generic models can generate code, but they do not know whether a bank's transaction approval flow, a healthcare compliance step, or an automotive design constraint applies unless that context is encoded and connected to the agent.
# Every architectural layer matters
Somasundaram emphasized that agentic capabilities cannot be confined to the application layer. Enterprise architecture is an interdependent stack: infrastructure, networking, identity and access, development practices, governance and security. Agents need to reason across that stack rather than only generating functional code.
Identity and access management are especially important. When agents exchange data, a hallucinated or incorrect assertion can propagate across agents and amplify into a faulty workflow. To prevent that, guardrails must exist at multiple layers: connectivity rules, authentication, SDLC controls, regulatory checks and security policies should be integrated into agentic workflows rather than applied as after-the-fact controls.
# How the developer role evolves
As agents take on more execution, human roles shift toward defining objectives, supplying business context, and validating outputs. Developers will increasingly act as context engineers: curating domain knowledge, encoding organizational constraints, and building the governance that prevents unsafe or incorrect agent behavior.
Somasundaram framed this as a change in the nature of engineering rather than a replacement of engineers.
# The operational challenge: testing and complexity
He referenced a looming concern labeled "Mythos" tied to the added complexity of agentic workflows and the need for enterprises to slow down and design for safe, verifiable outcomes.
# Practical implications
The promise of neurosymbolic AI — which combines neural network pattern recognition with rule-based reasoning — will only be possible when underpinned by trusted, governed business context. Context has become a buzzword, with terms like semantics, ontology, semantic layer, knowledge graph, and context layer being used
Cisco’s MyAgent moves AI beyond chat, delivering secure, personalized ambient intelligence with persistent context that turns intent into action.
As 2025 was the year GenAI took hold in RegTech solutions, 2026 is seeing the rise of AI agent adoption raising the upside for compliance automation and creating new risk surfaces. Early GenAI solutions enabled unstructured communications, policies and case files to combine with structured transaction and customer data
Atlassian today extended the capabilities of its portfolio to provide teams of artificial intelligence (AI) coding agents with the level of context needed to build and deploy applications in production environments at much higher levels of scale. Code Context, built on Atlassian’s Teamwork Graph, makes it possible for
As AI matures, enterprises and customers are rapidly deploying agents seeking to unlock the next level of automation and productivity. Agentic AI shows potential to handle a multitude of use cases, from buying personal items on Amazon to customer service applications to enterprise security and software development. How
Cybersecurity spent decades trying to stop things from happening, identifying them, and then investigating what happened. Agentic AI challenges us with a much more difficult question: What was the system trying to do? Agents don’t just execute instructions. They interpret objectives, select tools, access data, cross sy
Loading more related stories...
Open the app view to save this story, compare related coverage, and continue from the same source.