Today's corporate leaders face an unprecedented mandate: steering organizations through entirely uncharted economic waters. As artificial intelligence reshapes the global marketplace, C-suite executives are paying close attention to how these emerging technologies translate into measurable business value.

Moving From Sandboxing to Production

During a panel discussion at the Imagination in Action event titled The Next Endeavor, industry leaders gathered to share strategic imperatives for modern enterprises. Moderated by Ankur Gaba of Deloitte, the conversation featured Rama Akkiraju of NVIDIA, Srini Namineni of Cisco, Satya Jaydev of SanDisk, and Nishit Sahay of Marvell Technology, offering a comprehensive look at current corporate strategies.

Gaba opened the discussion by highlighting industry estimates showing that roughly 25% of organizations have successfully transitioned about 40% of their artificial intelligence use cases out of the sandbox and into active production.

Where the Big Gains Happen

When discussing primary areas of return, NVIDIA's Rama Akkiraju pointed directly to chip design and verification testing as core operations heavily augmented by AI. She noted that their software development is entirely powered by coding tools like Claude Code, Codex, and Cursor. The entire enterprise ecosystem and data access systems have been integrated using model context protocols, allowing all company engineers to code using these advanced frameworks.

Akkirajy also pointed to supply chain improvements:

  • Demand and supply planning
  • Scenario analysis
  • Cybersecurity and code vulnerability assessments
  • Enterprise-wide autonomous agents across IT, employee productivity, and digital assistance teams

Similarly, Srini Namineni reported success across multiple business domains at Cisco, emphasizing that the software development life cycle within research and development leads the charge, closely followed by sales, human resources, finance, and legal operations.

Just three weeks ago, we actually have launched a personalized agent for every single employee in the company. What it means is they don't even have to go to an assistant and ask questions anymore. They can give standing instructions, and it'll connect to Outlook or calendars, as well as many other MCP servers.

Srini Namineni, Cisco

Satya Jaydev of SanDisk described his organization as standing at an inflection point regarding its artificial intelligence adoption, utilizing a three-pronged strategy that spans productivity use cases, disruption scenarios, and design automation.

Clarifying Infrastructure and Agentic Flows

Nishit Sahay offered a clear definition of Marvell Technology's position in the ecosystem, noting that while GPUs serve as the brain of AI, his company builds the nervous system that connects these processors into massive systems capable of handling inference, agentic flows, and model training.

According to Sahay, product development remains the north star for agentic workflows, but fundamental operational readiness is critical. Achieving true return on investment requires understanding existing workflows, foundational model setups, data pipelines, and safe automation processes.

The Leap to Autonomous Agents

Addressing the most surprising shifts in the current landscape, Akkiraju emphasized the leap from simple query-answering systems to autonomous agents capable of executing complete end-to-end tasks. Ensuring these agents operate securely without unbounded autonomy requires rigorous testing environments, virtual machine sandboxes, and Kubernetes-level isolations.

Panelists also addressed unexpected challenges:

  • Token Costs: Token expenditure accelerated much faster than anticipated, forcing companies to optimize on-premise infrastructure.
  • Probabilistic Nature: Large language models are inherently nondeterministic and prone to hallucinations, keeping humans firmly in the loop.
  • Technical Debt: Organizational technical debt becomes vastly magnified under accelerated tech strategies, acting as a heavy drag on resources.

Sahay concluded with a stark reminder regarding security and financial impact, noting that empowering an organization with full autonomy immediately brings long-ignored operational flaws to the surface, making robust security protocols and meaningful tokenomics vital for long-term success.