Personal agents kept coming up at the NiCE Analyst Summit I attended last week. What happens when customers can hand the effort of getting customer service over to an AI agent like Meta Muse or Instinct? And what happens when that agent is willing to keep trying long after most of us would’ve given up?
NiCE executives expect personal agents to place increased demand on contact centers. It’s easy to see why. There are plenty of things we don’t pursue because they take too much time. We might decide a billing discrepancy isn’t worth a phone call or put off asking about a warranty until we’ve forgotten about it. Give those tasks to a capable personal agent, and more of them could become service requests.
That could be good for consumers. But companies will need to prepare for the additional work.
A customer who keeps trying
One concern raised at the summit was that persistent agents could generate traffic resembling a denial of service attack. An agent trying to resolve a legitimate problem might keep calling, retry a failed request or pursue several channels at once. Multiply that behavior across thousands of customers, and the operational implications become significant.
The response NiCE executives emphasized was to automate as many of these customer intents as possible and make sure the infrastructure can scale. If personal agents can get things done efficiently, there’s less reason for them to keep coming back.
I think that last point deserves attention. Adding capacity will help, but companies will also need to look at what causes repeated contacts. An automated system that answers a question without completing the requested action could leave a persistent personal agent trying again and again. At the summit, we saw how NiCE uses LLMs to review customer interactions and identify opportunities to improve processes and automate tasks. It can even build AI agents for people to review and publish.
Defining the terms
Authentication and verification were acknowledged at the summit, although they received less attention than demand and automation. They’ll be essential as agents take on tasks involving account information, purchases, refunds and changes to services.
Work on the emerging Personal Agent Protocol reflects that need. Meta and Sierra announced the initiative with several industry partners, and Phil Heltewig, Chief AI Officer, NiCE / Co-Founder Cognigy, confirmed during his presentation that NiCE is participating.
The effort aims to give personal agents and businesses a consistent way to interact, with consumer permissions and business controls built into the relationship. There are critical details to establish about whom an agent represents, what it’s authorized to do and what access a business should provide.
Getting those details right will help protect consumers, merchants and enterprises. The protocol is still under development, so we’ll need to see how the specifications, implementation and adoption play out.
From shopping to service
Our new Opus Research report, Agentic Commerce: A 15-Year View of a 2-Year-Old Category, provides some background for understanding how we got here.
For years, conversational commerce promised that we’d be able to tell a system what we needed and have it help us get things done. Today’s AI agents bring much more of that promise within reach, with the ability to research options, compare products and take action on a shopper’s behalf.
The report explores that evolution, the different kinds of agents entering the market and what increasingly capable shopping agents mean for consumers and brands. It examines who these agents work for, how much authority consumers are willing to delegate and what businesses need to prepare for.
The summit discussion points to the next part of the story. A personal agent that helps someone buy a product may later be asked to arrange a return, challenge a charge or pursue a warranty claim. Those interactions have particularly significant implications for CCaaS and CX vendors, reaching directly into the systems and processes they provide.
We’ll explore that territory more closely in a separate upcoming report on agentic service. There’s plenty to work through, and last week’s discussions made clear that it’s already on the industry’s agenda.
[Download the Opus Research Agentic Commerce report.]
Categories: Articles
2026 BPO, AI & CX Intelliview: From Labor Arbitrage to Managed AI Operations
How AI Agents Are Changing Shopping and What Comes Next
Invoca Wants to Turn Google Searches into Customer Relationships
Dialpad and Ada Combine to Build a Platform for Customer Work