Zendesk unveils Outcome AI Platform to adapt customer service to personal AI assistants

Zendesk unveils Outcome AI Platform to adapt customer service to personal AI assistants

Zendesk presented the Outcome AI Platform at its AI Summit. Six months after unveiling the Autonomous Service Workforce at Relate, the vendor is reorganizing its offering around a shift in usage: more and more consumers are entrusting their tasks to personal AI assistants.

According to a global survey conducted by Zendesk, 52% of consumers surveyed are comfortable with the idea of relying on these assistants for everyday tasks, and 63% believe companies will need to rethink their customer service to accommodate them. However, they remain attached to human interaction: 84% believe human assistance should always remain accessible. The platform assembles four components and comes with a new pricing offer. Tom Eggemeier, CEO of Zendesk, summarizes the logic: ‘People don’t ask which channels, systems, or services they need to go through to interact with a company. They simply want help accomplishing what they came to do.’

An open directory to direct assistants to brands’ official support

To act on behalf of a consumer, an AI assistant must first know who to contact. This is the goal of the Open Service Directory, which combines an open specification with a public directory. Assistants can find the verified support channel of an organization there, whether it uses Zendesk or not. Each brand registers its channels, chooses the information it shares, and sets the cases where human validation is required. The directory does not access any private customer data and does not perform any service actions.

Among the first participants are Betterment, ElevenLabs, Flex, TeamSystem, TBA Auctions, Telekom Payment S.A., and Wireless Logic. Michael Chaouat, VP Sales Southern Europe at Zendesk, describes the French market: ‘We see growing interest in France, even though demand remains emerging and the majority of companies have not yet deployed dedicated channels.’ He cites Zendesk’s CX Trends 2026 report, which states that 85% of French CX leaders say the tangible value generated by AI has exceeded their initial expectations.

According to Zendesk, access to the directory is free and does not require being a customer of the vendor. A brand’s journey involves three steps: a request submitted on the dedicated portal; once validated, integration of the verified destination into the existing service ecosystem; and then end-to-end testing with partner assistants. According to the vendor, these tests aim to ensure routing accuracy and data security before public opening.

Zendesk Bridge, Agent Builder, and agents gaining autonomy in stages

Zendesk Bridge does the reverse of the platform: it brings Zendesk’s capabilities into the tools teams already use. The component connects internally developed agents and assistants used by employees, such as ChatGPT or Copilot, to Zendesk’s AI, its service context, and its workflows. Each action goes through Bridge, which applies identity and permission controls, rules defined by the company, human validations, as well as observability and audit.

At Relate, Zendesk had presented Agent Builder, a no-code interface for designing agents tailored to each company’s policies. Specialized agents, which include Industry Agents and Custom Agents, now tackle complex, multi-step tasks. Michael Chaouat believes the main obstacles lie in ‘data fragmentation, siloed systems, and ambiguity about responsibilities.’ He adds: ‘Autonomy is not a binary switch: it extends gradually at the pace where AI demonstrates concrete and reproducible business results.’

Zendesk Unlimited replaces per-seat billing with an outcome budget

Zendesk already bills its AI agents per verified resolution. With Zendesk Unlimited, user access becomes unlimited and per-seat billing, i.e., per license assigned to each user, disappears. Each customer sets an Outcome Budget and only pays for eligible resolved requests and completed business work. The vendor claims the platform can automate up to 84% of service requests and reduce resolution times from several hours to a few minutes.

Tapan Patel, research director at IDC, believes linking AI investment to verified resolutions rather than the number of seats aligns cost with value obtained. The platform components are not all arriving at the same pace: specialized agents and Self-Improving Zendesk are generally available, Zendesk Bridge is offered in early access, and the Open Service Directory will be launched soon.

S’abonner
Notification pour
guest

0 Commentaires
Le plus ancien
Le plus récent Le plus populaire
0
Nous aimerions avoir votre avis, veuillez laisser un commentaire.x