Consumer-facing AI agents are now handling subscription cancellations, hold-time management, and dispute resolution on behalf of customers, creating a direct challenge to traditional support operations. This shift reflects a broader market recognition that routine transactional work—particularly the friction points customers actively avoid—represents a viable target for automation. The emergence of purpose-built tools like Ema, which raised $77M on the premise that AI can displace enterprise software workflows, signals that this isn't a niche experiment but a scaling trend. For support teams, this creates an immediate tension: these third-party agents operate outside your CRM, bypass your routing logic, and interact with your systems as adversarial actors rather than integrated partners. The question becomes whether your current infrastructure—whether Zendesk, Freshdesk, or Salesforce—can even detect and appropriately handle requests originating from non-human actors, let alone route them to the right queue or preserve customer context.
The strategic implications cut deeper than operational friction. If customers increasingly delegate their own service interactions to AI intermediaries, support teams lose direct visibility into customer intent, frustration triggers, and the actual reasons behind cancellations or complaints. This inverts the traditional CX data flow: instead of your team gathering signals to improve retention and experience, you're now receiving sanitised, agent-mediated requests stripped of emotional context and negotiation opportunity. Simultaneously, as agents start acting for brands, the question of accountability becomes acute—when a third-party AI misrepresents your policies or makes commitments on your behalf, who owns that failure? For teams already running native AI solutions like Agentforce, this raises a critical competitive question: does your platform's value proposition depend on being the *only* AI agent in the conversation, or can it coexist with customer-side automation? The risk is that support becomes a battleground between opposing AI systems rather than a space for human judgment and relationship repair.
The longer-term concern is structural. If routine work genuinely migrates to customer-deployed agents, support teams will be left handling only the exceptions—complaints, escalations, and edge cases—without the volume or predictability needed to justify current staffing models or training investments. This could accelerate the shift toward smaller, higher-skilled teams, but only for organisations that can afford the transition. Mid-market and smaller support operations may find themselves caught between rising complexity and shrinking transaction volume, unable to justify the investment in AI-native platforms whilst losing the efficiency gains that routine work once provided.
People are sending AI bots to sit on hold, cancel subscriptions, and fight customer-service battles for them Business Insider