The future of service belongs to self-improving AI

We’re at a watershed moment in service, one where systems of record give way to intelligent systems of action. Static automation becomes not only agentic, but self-learning, and customer outcomes improve with every interaction.

For years, we’ve been building toward this point: a future where AI capabilities can finally match the demands of real-world service. Where AI isn’t just assistive, but agentic and capable of true end-to-end resolution, even for issues that are multi-step and complex.

Today, we’re taking a decisive step forward. By signing a definitive agreement to acquire Forethought, we are advancing our vision for the world’s most innovative agentic service business and extending agentic AI capabilities to any service team.

AI that improves with every interaction

Zendesk AI customers are routinely achieving 80%+ automation rates, and we expect that AI agents will handle more service interactions than humans by the end of the year. As AI takes on this greater share of responsibility, however, it must meet a higher service standard.

Quality and consistency are paramount, as is the ability to resolve complex multi-step issues end to end. Agentic AI can already work across systems, take actions even when no APIs are available, and carry issues through to full resolution, but it requires heavy manual oversight to maintain. Teams must tune prompts and continuously update workflows as new scenarios emerge.

With the Zendesk Resolution Learning Loop built in, every interaction becomes a self-learning input. Teams don’t have to become prompting experts; complexity and volume are what train the system, and every resolution becomes a new baseline upon which to improve.

Unlock the power of Voice AI

This opportunity for self-improvement will now extend to voice. There will always be moments where customers just want to pick up the phone, yet voice has historically been one of the hardest channels to scale and learn from.

Voice AI changes this dynamic, turning a previously siloed channel into a scalable one. With our proposed Forethought acquisition, unstructured speech will become actionable intelligence that will make it possible for Voice AI agents to understand and navigate complicated issues such as processing a refund mid-call.

When needed, these agents will be able to seamlessly hand off to their human counterparts with the full context needed to avoid any disruption in service. And just like with other channels, the system will use any learnings from tone of voice or other customer sentiment markers, once nearly impossible to capture at scale, to continuously improve.

Zendesk AI on any service platform

Until now, the power of Zendesk AI agents could only be leveraged through our own service platform. Once the Forethought acquisition closes, however, we will be extending Forethought’s capabilities to make agentic, self-improving systems available to all service teams.

While these agents work best within Zendesk, progress toward agentic service should not be limited by the systems teams use today. By extending these capabilities more broadly, we will be enabling more organizations to move toward a model in which resolution compounds and quality improves under load, while giving teams the flexibility to evolve at their own pace.

Deliver the new service standard

Customers don’t care how efficient you are at deflection or how much you’ve reduced internal costs. They want their questions answered and their issues quickly and meaningfully resolved.

Historically, achieving that level of service required linear increases in headcount. Today, self-improving, agentic systems make it possible to consistently meet this standard, even in the face of growing demand.

These self-learning systems are the future of service, a future we intend to lead.