Blog

Getting AI right in customer support from the CX leader perspective

For CX leaders, most questions being asked about AI are probably about technology: which platform to choose and how much of the volume it can take on? However, much of the outcome depends on two decisions around the technology. The first one is what your contract and dashboard measure; the second is what your customers see before they decide to trust an agent.

Cropped photo of a young customer service agent speaking with customer with headphones and in front of computer.

For CX leaders, most questions being asked about AI are probably about technology: which platform to choose and how much of the volume it can take on? However, much of the outcome depends on two decisions around the technology. The first one is what your contract and dashboard measure; the second is what your customers see before they decide to trust an agent.

This month’s reading looks at that from the CX leader’s seat: what buyers are seeing in their outsourcing contracts and AI programmes, and what we know customers need before they trust an AI agent for CX.

What’s worth your attention this month?

1. If the client is unhappy when every SLA is being met and exceeded, sometimes the problem could be what is being measured in the first place

In The BPO Industry Says It Wants AI - The Contracts Say Otherwise, Mark Hillary summarises his conversation with Victor Manzanera, founder of Scale Edge Group.  Manzanera describes a situation many buyers will recognise: the client is unhappy with the relationship, and the performance dashboard shows every contractual KPI as green. His review of more than 50 outsourcing contracts from 2017 to 2025 found no bonuses tied directly to business outcomes, and around 70% of incentives were attached to lower-level operational metrics. His practical advice: decide which outcome the business wants, such as retention or first-contact resolution, and tier the KPIs so that bonuses sit against it.

Read The BPO Industry Says It Wants AI - The Contracts Say Otherwise by Mark Hillary

2. Two-thirds of buyers expect their outsourcing provider to lead on AI, and only 21% say they have the skills to govern it

ISG’s State of BPO Report, based on a survey of 250 senior decision-makers, shows where buyers stand. 70% of respondents expect their BPO providers to deliver more innovation, and more than half have already moved beyond manual or labour-led BPO models. Yet, AI is still used mainly for efficiency (43%), and only 14% report AI-first outcomes.  Also, less than one quarter say they have the internal skills to govern AI-enabled services, and delivery of AI-led innovation is the lowest-rated return on investment as well as the least measured. ISG principal analyst Michael Dornan, the report’s co-author, says providers that can measure automation impact and stay accountable to industry-specific KPIs are better placed to prove value and expand client relationships.

Read AI Redefines BPO Success as Innovation, Outcomes: ISG Study

3. Your customers are warming to AI agents, and helping them trust one is within your control

Fin’s 2026 AI Sentiment Report, based on a survey of 1,026 US end users, argues that most businesses are earning trust in AI agents by accident. Sentiment is warming, with 49% describing an AI agent interaction for customer service as positive, up 9% from 2025. That figure rose to 74% after they watched an agent resolve a real query. Trust still varies by task, with 62% willing to trust an agent to check an order status without a human, and only 20% would for help during a travel disruption. When people call a query too complex for AI, they mostly mean it needs caring and emotional awareness as well as computing. Asked what would make them more comfortable, 65% chose easy access to a human.

Fin’s own support team reports that one in five queries reaching its human team carries a trust signal, where the customer asked for a person before letting the agent try or did not believe an answer that was right from the start.

Read The 2026 AI Sentiment Report by Fin

What do we think about it?

1. An AI agent's early resolution numbers can look great while still leaving half your customers unserved

In “How to Improve AI Agent Performance After Progress Plateaus”, Lorcan McGeough, CX Optimization Specialist at Otonomee, looks at what happens once the early gains from an AI agent have been captured. Most deployments resolve close to 90% of simple, informational queries early on and then stall as the agent meets more complex or unpredictable requests. Lorcan sets out five fixes: know exactly which queries you are dealing with and how complex each one is, put governance of your knowledge base in place before you scale, map the process end to end, build the business case with real volume and value data, and measure involvement rate, resolution rate and automation rate separately, since a dashboard showing an 80% resolution rate can still mean half your customers never reach the agent at all.

“People often confuse these rates, and then end up trying to fix the wrong thing.” Lorcan McGeough, CX Optimization Specialist, Otonomee

Read How to Improve AI Agent Performance After Progress Plateaus

2. AI in customer support isn't being framed as a cost play in Silicon Valley right now, it's being framed as a growth one

Aidan O'Shea, Otonomee’s co-founder and CEO, shared his reflections after two weeks in San Francisco attending Pioneer by Fin, while also meeting customers on his trip. He noted that the mood was positive, with the discussion focused on opportunity and growth, and with AI seen as something that augments people. In customer support, AI is being discussed as a once-in-a-lifetime chance to reach more customers and increase their lifetime value. It lets companies scale efficiently and move budget towards demand that previously went unmet, which supports sustainable revenue growth.

“It feels like there is an urgency for companies to lead in their space and capitalise on the possibilities created by AI.” Aidan O’Shea, Co-Founder and CEO, Otonomee

Where we’ve been

1. Pioneer by Fin, San Francisco

Otonomee was a partner sponsor at Pioneer, Fin’s annual conference, where Aidan and Hilary O'Shea joined our Chief Commercial Officer, Alan Gleeson, alongside Fin’s leadership and other companies working on AI agents. Fin reported at the event that it is now resolving over 10 million conversations a month at an average resolution rate of 76 percent, and interaction volumes are likely to keep rising as AI makes it easier to contact a business. That leaves human agents with the more complex and more valuable work. One idea from Fin co-founder Des Traynor’s keynote stuck with the team: “We shape our tools, and after that our tools shape us”.

Read more about Fin Pioneer

2. Sunday Business Post: Aidan O’Shea on taking the simple work out of customer support

Aidan spoke to Emmet Ryan of the Sunday Business Post about how the outsourcing role is changing as AI takes on more of the simple, repetitive work. His view is that AI dropped into a badly run operation tends to make it worse, so businesses need a clear view of their customer journey and escalation routes before deploying AI. The work left for human agents becomes more demanding and valuable, which he sees as good news for locations such as Ireland, where people need to add value at the complex end of support rather than the basic one.

Read the full feature

Closing thoughts

The thread through this month’s reading is that the results of an AI implementation depend on decisions that sit around the technology. A set of KPIs can report green when the business needs something else, and a high-resolution rate can leave out the customers who never reached the AI agent. For CX leaders reviewing an AI plan or an outsourcing partner, a useful question is what the current metrics would report if everything worked exactly as intended, and whether that is the result the business wants.

If outsourcing is part of how you are thinking about your support model, we would be glad to be part of that conversation.

Get in touch

‍

More recent news

Cropped photo of a young customer service agent speaking with customer with headphones and in front of computer.
Otonomee Insights
5 Nov
,
2026
16 Nov
,
2026

Getting AI right in customer support from the CX leader perspective

Getting AI right in customer support from the CX leader perspective
Woman sitting at home desk wearing headphones smiling
CX Optimization
29 Oct
,
2026
30 Oct
,
2026

How to Improve AI Agent Performance After Progress Plateaus

How to Improve AI Agent Performance After Progress Plateaus
Pioneer by Fin event banner with image of speaker on stage
Customer Experience
Events
16 Oct
,
2026
19 Oct
,
2026

Otonomee Fin Pioneer

Otonomee Fin Pioneer