How far should AI go in customer support and success?
When it comes up
When support costs rise, when a vendor proposes a customer agent, or when renewal teams are stretched.
Your options
AI assists human agents and CSMs
Fits when: Complex products, high-value accounts, or customers who expect a person.
Watch out: Productivity gains depend on training and workflow; measure them.
AI agent for routine questions with human hand-off
Fits when: High volumes of routine questions with clear, documented answers.
Watch out: Accountability for wrong answers, hand-off speed, customer preference for humans.
AI-first service with minimal humans
Fits when: Very simple, low-value interactions where customers prefer speed.
Watch out: Quality and retention risk; public backlash if customers can't reach a person.
How to decide
Segment interactions by value and complexity. Use agents for routine, well-documented questions; keep humans for complex, sensitive and high-value accounts; use AI to make humans faster everywhere. Measure retention, satisfaction, hand-off time and complaints — not only deflection — and decide what the agent may commit to.
Trade-offs
Agents lower cost per contact and give instant answers but create accountability and preference risks; human-led service protects relationships but costs more; assistive AI is lower risk but its gains are smaller per contact.
Real examples
- In Moffatt v. Air Canada (2024 BCCRT 149), a tribunal held Air Canada responsible for wrong information its website chatbot gave a customer, rejecting the idea that the chatbot was responsible for its own actions. Source: McCarthy Tétrault — Moffatt v. Air Canada: a misrepresentation by an AI chatbot (2024 BCCRT 149) ↗
- CX Today reported that HubSpot's Customer Agent resolved 70% of support conversations on average in Q1 2026, up from 20% a year earlier, across more than 9,000 customers. Source: CX Today — HubSpot's Customer Agent hits 70% resolution rate in 12 months (May 12, 2026) ↗
- Klarna's CEO said in May 2025 that 'From a brand perspective, a company perspective, I just think it's so critical that you are clear to your customer that there will always be a human if you want', as the company resumed hiring human agents. Source: Entrepreneur — Klarna is hiring customer service agents after AI couldn't cut it on calls, according to the CEO (May 9, 2025, citing Bloomberg) ↗
What to check first
- Which question types make up most volume, and whether approved answers exist for them.
- What the agent would be allowed to commit to (refunds, credits, changes) — ideally nothing without rules.
- Retention and satisfaction baselines for the accounts the agent will serve.
Learn and use
Sources
- McCarthy Tétrault — Moffatt v. Air Canada: a misrepresentation by an AI chatbot (2024 BCCRT 149) ↗
- CX Today — HubSpot's Customer Agent hits 70% resolution rate in 12 months (May 12, 2026) ↗
- Entrepreneur — Klarna is hiring customer service agents after AI couldn't cut it on calls, according to the CEO (May 9, 2025, citing Bloomberg) ↗
- Gartner — Survey finds 64% of customers would prefer that companies didn't use AI for customer service (Jul 9, 2024) ↗
Last reviewed 2026-10-01.