Communications service providers expect the majority of customer service interactions to be handled by artificial intelligence (AI) within the next two years, but there may be some gaps to fill before consumers trust the technology to deliver, based on research published by Amdocs.
Based in Jersey City, NJ., where it provides a range of software platforms and services to the CSP sector, Amdocs commissioned a survey of approximately 7,000 consumers and 130 telecom decision-makers across 14 countries to produce its second annual report, titled Agent Personality Engineering: From Vision to Value.
Amdocs found that 77 per cent of consumers already have a baseline understanding of AI and 69 per cent said highly effective AI agents would positively affect their perception of a service provider. That’s a nine per cent jump since Amdocs fielded the same question last year.
On the other hand, the proportion of consumers who doubt AI’s follow-through also rose eight per cent. from 2025. Not only are consumers concerned the resolution to their problem that emerges from chatting with an AI agent won’t be implemented, but more than half (52 per cent) worry they won’t be able to reach a human service rep if they need one.
Consumers also have high expectations of AI agents around retaining context: 72 per cent said they expect the technology to ensure continuity across interactions, and 59 per cent said they want “mood-adapted” responses. Even more (61 per cent) want AI agents that demonstrate empathy in some way.
“In the agentic Era, every AI-led interaction is a brand moment,” Amdocs said in a statement accompanying the research. “Brands must orchestrate these AI personas with the same rigor and empathy they expect from their best human talent. Achieving true customer intimacy at scale is no longer just a marketing goal; it is a growth imperative.”
360 Magazine
If the 24-month timeline for heavily AI-powered customer service sounds a little ambitious, it should. It definitely doesn’t line up with what CSPs are actually doing with the technology so far.
For example, Amdocs found 84 per cent of projects so far are limited to basic “copilots” where (probably generative AI) is used alongside human reps. This may not be the maturity gap that Amdocs suggests so much as a prudently cautious approach to evaluating the technology before it is deployed at scale.
One lingering question that may be included among the more technical concerns is whether an AI agent should have a distinctive brand personality. Just over half (52 per cent) of CSPs say this is on their to-do list within the next one to two years, but it may take longer for them to fine-tune the way it’s rolled out and what level of human oversights is needed once customers use them.
Personalizing AI agents is clearly an area where Amdoc is ready to help: it showed doing so led to a 3x boost in customer loyalty and likelihood to recommend a service provider. This was in testing scenarios, however, rather than a real-life case study.
Oddly, 56 per cent of CSPs said they would have multiple predefined personalities for AI agents over the next through years. You can applaud the intention to tailor experiences towards specific segments, but the more critical piece is getting AI agents to a place where the bulk of customers accept them and don’t constantly ask to speak to a human being.
The Amdocs statistics were included in a press release but the full report doesn’t appear to be online or in its Resource area, at least at press time.