Scaling Sustainably with AI in 2026: Striking the Right Balance

In conversation with Michelle Donnelly, Chief Revenue Officer at Crescendo AI

Right now, the AI market is saturated with demos, there’s simply no avoiding them. Yet, no matter how impressive they may be, the question CX leaders need to be asking is how to strike the right human / technology balance when building true, connected customer experiences.

We sat down with Michelle Donnelly, Chief Revenue Officer at Crescendo AI to break down that balance and pinpoint where these demos are falling short.

Through her experiences across Salesforce, Groq, and Crescendo, Michelle gave us some real insight into how to scale sustainably and responsibly, all while building repeatable customer motion.

Are We Ready For AI?

Our latest Call Centre Helper research, as well as wider industry trends, showcase the gap between AI eagerness and readiness.

Put simply, it’s easy for a lot of AI companies to look great in demos, but can we be sure this will translate to real customer environments?

The fact is, demos are tightly controlled, customer interactions are not.

During our conversation, Michelle painted a familiar picture: imagine you are a retail leader, looking to expand your CX tools with technology you find at a retail show.

These shows tend to be noisy, colourful and entirely saturated with demos that are all designed to shine in this one specific environment.

It’s easy to get swept up in the pageantry to later find yourself with something that winds up not fulfilling the true needs of your business.

So, what is it that these demos are missing? “The thing that nobody was talking about,” Michelle explained, “was the value of the human in the loop.”

Keeping people in the conversation, the importance of escalation

An AI that ignores the value of human input can only go so far. In order to scale sustainably, it’s imperative we include the human agent in the conversation.

For instance, where some companies focus only on implementing effective AI agents in the front end, Crescendo takes it a step further and seamlessly hands off situations to humans when the need arises.

By incorporating humans in the process, it allows for a better customer experience. We’ve all been in the situation when we need a question answered and the AI chatbot is simply not getting the job done.

This is where the human agent should step in, refocus the conversation and get the customer the answer they’re looking for.

Employing AI With Purpose

When buying into a new CX tool, the aim is always the same: to ensure your customer (and your customer’s customer) gets what they want. To do that, you have to be sure you’re really getting what you’re buying, and that can’t be done with demos alone.

During the leadup to the Super Bowl, Michelle described being approached by a betting company that was looking for advice on how to handle the pure scale of customer interaction on the horizon.

24 hours after their first conversation, Crescendo presented them not with a demo, but with an actual AI agent that was built for their direct purpose, assisting in customer experience on their website.

“Through scraping the public knowledge base, AI agents like this can be created with a clear purpose and the ability to scale exponentially.”

By implementing AI to scrape databases, more information can be gathered in minutes than a human agent could gather in a week, allowing more time for the human agent to focus on the questions that only they can answer.

Both Human & Machine Learning

Asking yourself the question of ‘how to best implement AI with existing employees?’ can raise reasonable concerns. With the right guardrails and AI tool, you can answer your own question.

Using AI to scrape your knowledge database allows human agents to train quickly and effectively, supporting them to access the information provided by the AI.

Human agents can build from this and take it a step further, tackling the more complex customer experiences that only a human touch can provide and leaving the AI to handle the ‘nitty gritty’.

Execution Speed vs Customer Trust

Striking the right balance between speed and accuracy is critical to any CX organisation. That’s why it’s essential we put the correct guardrails in place.

Think back to when ChatGPT first came out. Every headline was centred on the idea that AI was coming to replace us. Even outside the customer contact space, from accountants to marketers, we would all soon be out of a job. And yet, here we are.

Rather than replacing us, the pendulum is almost swinging back the other way. Everyone understands the power of AI and the value it can add to the CX process, but AI is just scraping the knowledge base. Everything it is built on is driven by humans.

Human agents remain vital to control what is being put into, and in turn put out of, each AI system, ensuring that customers are getting what they truly need.

The human in the loop approach is the key to maintaining quality and scaling not just for the sake of it, but with purpose.

As Michelle lays out, when building an AI agent, you can’t just “set and forget”. You have to work with it, incorporate it into your specific environment and feed it accurate data that is relevant to your situation.

AI will only continue to improve if the right kind of quality control is put in place and human eyes are there to monitor and course correct as needed. The human touch is vital to driving insights from AI and determining how best to use the information that is produced.

AI technology is constantly changing – even in the last few months, we can see rapid growth across the CX field. But this growth will only be sustainable if we coax it with a human hand and help to mould it into what our customers are looking for.

Remember, it’s that human touch that will set you apart and have the customer coming back each time.

Author: Hannah Swankie
Reviewed by: Xander Freeman