NiCE COO Arun Chandra on scaling agentic AI past the pilot stage
This is a sponsored article: MIT Technology Review's custom content arm, Insights, produced it in partnership with NiCE, and it is framed as a webcast interview rather than independent reporting. The subject is NiCE chief operating officer Arun Chandra's view on how enterprises move agentic AI from pilots to company wide use. The piece opens with a figure: about 80% of Fortune 500 companies have adopted agentic AI in some form, but progress toward meaningful scale stays uneven, with many organizations still stuck running isolated pilots.
Chandra says the first step is moving past experimentation for its own sake. "Everybody's trying to figure out what can we do with this technology?" he says, but argues that scaling needs a clearer link to business strategy: organizations have to decide whether the goal is more revenue, lower costs, or another strategic or financial outcome, and then rethink the workflow an agent will run in rather than bolt AI onto what already exists. "The last thing you want to do is to apply AI on an outdated or an inefficient workflow," he says.
From there, Chandra frames agentic AI as something that has to be treated as one system rather than a collection of separate tools. Agents need access to the data, knowledge, and context to make good decisions, plus connections into back end systems if they are meant to act on those decisions; fragmented information undercuts both. "The efficacy of these AI agents is purely a function of the context, the knowledge, and the data they can ingest and use," he says.
On the organizational side, Chandra warns that scaling agents can create a new kind of fragmentation if different teams build systems that do not talk to each other, and that governance, privacy, security, and change management all become more important once agents take on more consequential work. He argues AI agents should eventually be held to the same standards as human employees, with companies treating their workforce as a mix of humans and AI agents. Looking further out, he suggests a connected setup could let agents work proactively and even communicate with each other to resolve customer requests. For companies moving from pilot to scale, his advice is not to try to "boil the ocean" but to build a connected strategy around a smaller set of high value use cases, workflows, workforce changes, and outcomes that can actually be measured.
Key facts
- This is sponsored content: produced by MIT Technology Review's Insights custom content arm in partnership with NiCE, not by the outlet's editorial staff.
- About 80% of Fortune 500 companies have adopted agentic AI in some form, but NiCE COO Arun Chandra says progress toward meaningful scale remains uneven, with many still running isolated pilots.
- Chandra says scaling requires picking a clear business goal, revenue growth, cost reduction, or another objective, and redesigning the workflow rather than applying AI to an existing, possibly inefficient process.
- He frames agents as needing unified data, context, and back end system access, warning that fragmented information and disconnected team level systems undercut their effectiveness.
- Chandra argues AI agents should eventually be held to the same standards as human workers and advises companies to target a focused set of high value use cases rather than trying to "boil the ocean."
Why it matters
Agentic AI pilots are common (NiCE cites roughly 80% Fortune 500 adoption in some form) but enterprise wide scaling is not, and this piece captures how a vendor executive frames that gap: not as a technology shortfall but as a strategy and workflow problem. The claim is that agents applied to unchanged, inefficient processes will not scale no matter how capable the underlying model is.
Who it affects
The audience is large enterprises already past initial agentic AI experiments and weighing how to move to broader deployment, plus the vendor, NiCE, whose COO is making the case in a piece it co-produced. No specific companies or industries besides NiCE are named as examples.
How to use it
The piece names no product, price, or specific tool from NiCE. What it offers instead is a set of principles from Chandra: define the business objective first, redesign the workflow before adding agents to it, give agents unified access to data and back end systems, and pick a narrow set of high value use cases rather than attempting broad rollout at once.
How solid is it
This is sponsored content commissioned with NiCE, not an independently reported story, and it says so explicitly. It carries one adoption statistic (80% of Fortune 500 companies) and no other figures: no cost, ROI, headcount, or timeline data, no named customer examples, and no date for when the interview took place. The claims are the perspective of one company's COO, not data NiCE or MIT Technology Review independently verified.
Risks and caveats
Read this as a vendor's framing of a market opportunity rather than neutral analysis: NiCE sells into exactly the agentic AI orchestration and governance problem Chandra describes. The advice given (align to strategy, redesign workflows, unify data access, start narrow) is broadly stated and not novel to this piece, and none of it is backed by NiCE specific case data in the source text.
“The efficacy of these AI agents is purely a function of the context, the knowledge, and the data they can ingest and use.”
— Arun Chandra, chief operating officer at NiCE