AI in Support — The 80/20 Reality CEOs Need to Plan For

The efficiency of AI in customer support is undeniable, but the future isn’t full automation - it’s a hybrid 80/20 model.

Industry research shows AI will soon be involved in virtually 100% of customer interactions, but it will only be able to resolve about 80% of them without human help (Zendesk). McKinsey estimates that up to 70% of customer contacts could be automated, but that leaves a meaningful share requiring skilled human judgment.

Salesforce’s own adoption highlights the limits: CEO Marc Benioff recently shared that AI now handles 85% of Salesforce’s customer service tasks with 93% accuracy, but the most sensitive cases still rely on people (Business Insider).

That final 20% matters disproportionately. These are the emotionally charged interactions, escalations, and complex cases that shape whether a customer stays loyal or defects.

At Sirius, we’ve built for this reality.

Our AI delivers scale, speed, and efficiency, while our highly trained human agents provide empathy, nuance, and problem-solving. For CEOs, the guidance is clear: don’t chase full automation. Instead, invest in the balance where AI handles efficiency and humans safeguard reputation.