Analyst(s): Keith Kirkpatrick
Publication Date: October 15, 2026
Zendesk has introduced the Open Service Directory, a draft specification and public directory that helps personal AI assistants find and connect to a brand’s verified support destination. The idea addresses a real trust gap in agent-to-agent service, but its success depends on assistant providers and rival CX platforms that have yet to commit to it.
What Is Covered in This Article:
- Zendesk announced the Open Service Directory (OSD), a specification and public directory of verified support destinations for AI assistants and personal agents.
- Brands opt in through their service provider, listings are verified against the brand’s own domain, and assistants connect to the brand’s approved support agent under the brand’s policies.
- OSD extends the Agentic Resource Discovery (ARD) specification and is designed to work alongside the UCP, PACT, and PAP communication protocols.
- Seven launch partners are helping shape the directory, which remains in draft with no pricing or general availability date.
- Futurum’s view on why adoption by AI assistant providers and competing CX platforms, not the specification itself, will determine whether OSD succeeds.
The News: Zendesk has introduced the Open Service Directory in a company blog post, describing it as a specification and public directory that helps AI assistants and personal agents find a brand’s official customer support and connect customers to it.
Brands opt in and approve a listing through their service provider, and any brand on Zendesk or another service platform can be listed. Listings are checked against the business’s own domain and screened for impersonation. An assistant searches participating registries, verifies the match, and connects to the brand’s approved support experience, where it operates within the brand’s authentication, authorization, and service policies. The directory itself does not access private customer data or perform support actions.
OSD extends the Agentic Resource Discovery (ARD) specification and complements the UCP, PACT, and PAP communication protocols, which Zendesk says it plans to adopt. Launch partners are Betterment, ElevenLabs, Flex, TeamSystem, TBAuctions, Telekom Payment S.A., and Wireless Logic. Zendesk says the directory is compatible with OpenAI’s Dots and ChatGPT, Claude, Muse, Instinct, Grok Bot, and Gemini, and that it is developing controls that let brands identify the assistant and the customer it represents, apply customer-specific permissions, and decide which providers can connect.
Will AI Assistants Use Zendesk’s Open Service Directory?
Analyst Take: Zendesk’s Open Service Directory tackles a problem that will grow quickly as consumers hand routine service tasks to personal AI assistants: before an assistant can reschedule a repair or request a replacement, it has to find the brand’s real support channel. By giving brands a verified listing that assistants can query, Zendesk is betting that discovery and trust, not conversation quality, will be the first point of failure in agent-to-agent service.
That diagnosis is correct. But the directory is the easy part. The hard part is getting the companies that build AI assistants and the CX platforms that compete with Zendesk to use a registry that Zendesk initiated. Until they do, OSD is a well-reasoned proposal, not a market standard.
Why the Open Service Directory Matters for Brands
The strategic risk for brands is disintermediation. If a personal assistant becomes the customer’s primary interface, the brand’s support experience can be bypassed, misrepresented, or reached through an unofficial channel, including the spoofed support numbers that already plague consumers.
OSD is designed to keep the brand’s own AI agent and service policies in the loop. The brand decides what data is exposed, which actions are allowed, and when a human takes over. For CX leaders, that preserves control over policy enforcement, escalation, and the customer record, which are what determine whether an interaction is actually resolved.
The handoff design also deserves credit. When human review is needed, the request moves to the support team with the conversation history and actions already taken. Context loss at handoff is one of the most persistent sources of customer frustration, and it gets worse when a third-party assistant sits in the middle of the exchange.
Open by Design, but Openness Alone Does Not Create a Standard
Zendesk chose not to make this a Zendesk-only directory. Any CX platform or AI provider can publish verified listings for the brands it serves, and the specification builds on ARD, an Apache 2.0-licensed discovery protocol already implemented by services such as GitHub’s Agent Finder and Hugging Face’s Discover.
That is the right call, and it is consistent with Futurum’s position that open orchestration must win out over proprietary agent ecosystems if enterprises are to avoid another generation of lock-in. A support directory only works if it is comprehensive, and no single CX vendor serves enough brands to make a proprietary registry useful to assistant builders.
Still, the incentives are not neutral. Zendesk gains influence by authoring the specification, and rivals such as Salesforce, ServiceNow, Genesys, and NICE have their own agent platforms and partner ecosystems to protect. If they back alternative discovery approaches, the result is a fragmented set of registries that recreates the very problem OSD is meant to solve.
Compatibility Is Not Commitment
The most important caveat sits in the fine print. Zendesk lists ChatGPT, Claude, Gemini, and other assistants as compatible, but the named launch partners are brands and platforms, not AI assistant providers. Compatibility means an assistant could use OSD; it does not mean OpenAI, Anthropic, or Google will query it by default.
The assistant providers also have reasons to build their own business-verification layers, since owning discovery gives them leverage over which brands and which service platforms their users reach. Without their participation, brands will be listed in OSD and wait for traffic that may not arrive.
Enterprises should also treat OSD as a direction rather than a deployable capability. The specification is in draft, the core documentation sections are not yet published, and the assistant identity and permission controls are still in development.
Outcomes and Measurement in Agent-to-Agent Service
Agent-to-agent service raises measurement questions that brands should start working through now. If a customer’s assistant negotiates with a brand’s AI agent, the brand needs to know whether the issue was resolved, how long it took, and whether the customer was satisfied, even though the customer never interacted with the brand directly.
This matters for pricing as well. As outcome-based pricing for AI agents becomes table stakes, vendors and buyers will need agreed definitions of what counts as a resolution when an external assistant initiates the request. Zendesk’s stated goal of visibility into who acted, with what permission, and what happened is the foundation for that accounting, but the controls must ship before brands can rely on them.
Ultimately, CX leaders should not wait for a standard to settle. They should audit where their support entry points are discoverable today, decide how they want to authenticate an assistant acting on a customer’s behalf, and define which actions they are willing to delegate. Those decisions will be required regardless of which discovery approach prevails, and brands that make them early will be ready to list in OSD, or anywhere else, when assistant traffic arrives.
What to Watch:
- Whether major AI assistant providers such as OpenAI, Anthropic, and Google formally commit to querying OSD, or build their own business-verification and discovery mechanisms that bypass it.
- Whether competing CX and contact center platforms, including Salesforce, ServiceNow, Genesys, and NICE, publish listings to OSD or back alternative discovery standards that fragment the market.
- Whether the assistant identity, permission, and audit controls Zendesk is expected to detail at Zendesk AI Summit 2026 are robust enough to satisfy regulated industries such as financial services and telecommunications.
- Whether launch partners such as Betterment, Telekom Payment S.A., and Wireless Logic report measurable gains in containment, resolution, or fraud reduction from verified assistant traffic.
See the complete announcement from Zendesk on its blog, including links to the developer documentation and the partner interest page.
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of Futurum as a whole.
Other Insights From Futurum:
Will Zendesk’s Forethought Acquisition Enable True Agentic Resolutions?
Has Agentic AI in Customer Service Finally Delivered on Its Promise?
Who Will Win the Agent Orchestration Layer Battle?
Author Information
Keith Kirkpatrick is VP & Research Director, Enterprise Software & Digital Workflows for The Futurum Group. Keith has over 25 years of experience in research, marketing, and consulting-based fields.
He has authored in-depth reports and market forecast studies covering artificial intelligence, biometrics, data analytics, robotics, high performance computing, and quantum computing, with a specific focus on the use of these technologies within large enterprise organizations and SMBs. He has also established strong working relationships with the international technology vendor community and is a frequent speaker at industry conferences and events.
In his career as a financial and technology journalist he has written for national and trade publications, including BusinessWeek, CNBC.com, Investment Dealers’ Digest, The Red Herring, The Communications of the ACM, and Mobile Computing & Communications, among others.
He is a member of the Association of Independent Information Professionals (AIIP).
Keith holds dual Bachelor of Arts degrees in Magazine Journalism and Sociology from Syracuse University.
