Zendesk
Also known as: Zendesk Suite, Zendesk Support, Zendesk AI agents, Ultimate.ai, Forethought
Customer service suite whose AI agents resolve across messaging, email, voice, API and web form using help-center knowledge, generative procedures or scripted dialogues, and authorized actions into connected systems.
Zendesk is a customer support suite with AI agents, knowledge management, omnichannel coverage and governance. Suite Team runs $55 per agent a month billed yearly, the Copilot add on is $50 per agent a month, and AI agents are billed per automated resolution.
Vendor details
Canonical URL
https://zendesk.com
Category
Customer support agent
Subcategory
Support — suite + AI agents
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
In practice
Your support runs on Zendesk already and bolting on a separate AI tool is friction. Zendesk brings AI agents into the suite, so automation rides on the platform you use.
Agents waste time hunting for answers buried in scattered articles. Zendesk pairs AI agents with knowledge management, so the right answer surfaces in the flow.
Enterprise support needs AI you can govern, not a loose bot. Zendesk includes governance across its AI agents and omnichannel coverage, so automation stays controlled.
Agentic Index coverage score
10.5 / 14 capabilities · 75%
| Integrations & Tool Calling | Full |
|---|---|
|
AI agents call authorized actions that read and write the customer's CRM, and API integrations that update data in any external system the customer specifies. The Apps framework and marketplace, Custom Objects API and Channel framework connect the wider suite to third party systems and telephony. Sourcesupport.zendesk.com/hc/en-us/articles/10783053154202read 2026-09-05 |
|
| Workflow Orchestration | Full |
|
Use cases identify what the customer is asking about and route each to either a generative procedure, which adapts the flow to the customer's responses while holding to business policy, or a dialogue, which runs defined branching logic. Both call authorized actions, API integrations to external systems and typed entities mid conversation. Sourcesupport.zendesk.com/hc/en-us/articles/8979864563610read 2026-09-05 |
|
| Knowledge Grounding & RAG | Full |
|
AI agents are bound to trusted knowledge sources, the Zendesk help center or an external knowledge base, which the platform indexes and draws on to generate answers, with guidance for optimizing that content for generative retrieval and generative search across it. The platform keeps that index independently of any conversation. Sourcesupport.zendesk.com/hc/en-us/articles/4408845739162read 2026-09-05 |
|
| Human Oversight & Guardrails | Full |
|
The customer defines exactly which actions an AI agent is authorized to perform, scopes each use case to either a flexible generative procedure or a prescriptive scripted dialogue with branching logic, customizes default system replies, and configures escalation to a human with full context. Action authorization and use case scoping are set before the agent acts. Sourcesupport.zendesk.com/hc/en-us/articles/10783053154202read 2026-09-05 |
|
| Security, Identity & Governance | Full |
|
Compliance covers SOC 2 Type II with annual audits, ISO 27001:2022, 27017, 27018 and 27701, ISO 42001 for AI management, FedRAMP LI-SaaS, PCI DSS, HIPAA with a BAA, HDS, Cyber Essentials Plus and CSA STAR AI. Customer controls include enterprise SSO over SAML, OIDC or JWT, native two factor authentication, IP restrictions, audit logs on Enterprise plans, and an add on for bring your own key encryption, retention policies and PII masking. Sourcezendesk.com/trust-centerread 2026-09-30 |
|
| Observability & Auditability | Partial |
|
The reporting dashboard covers automation metrics and AI agent outcomes, voice records and transcripts attach to their tickets, and conversation data is retrievable programmatically for analytics and compliance. There is no per conversation trace showing why the agent chose a procedure, action or knowledge source. Sourcesupport.zendesk.com/hc/en-us/articles/9510024609178read 2026-09-05 |
|
| Memory & State Persistence | Partial |
|
Developers can store and reuse parameters and metadata tied to each user or session, and entities capture typed values from a customer's messages within a conversation flow. Zendesk describes no lasting memory with a set scope or retention period. The Resolution Learning Loop improves Zendesk's own models, and customers cannot locate or delete what it learns. Sourcedeveloper.zendesk.com/documentation/ai-agentsread 2026-09-05 |
|
| Deployment & Data Residency | Full |
|
The Data Center Location add on lets a customer choose hosting in the United States, the European Economic Area, the United Kingdom, Japan or Australia. It is free of charge on Suite Professional and higher, purchasable on legacy plans, and covers Support, Chat, Guide, Gather, Explore, Talk and WFM. Some processing by Zendesk group members, their providers and non Zendesk services still falls outside the chosen region. Sourcesupport.zendesk.com/hc/en-us/articles/4408838409754-about-the-data-center-location-add-onread 2026-09-30 |
|
| Prebuilt Agents, Templates & Packs | Partial |
|
Out of the box bot automation ships ready to answer from connected knowledge sources with default system replies, and the developer surface offers conversation flow templates and message types. Agents are built from scratch, by selecting knowledge sources and then authoring use cases, procedures and dialogues, with no template gallery or prebuilt agent library at any step. The Marketplace carries apps and integrations, not agents a customer can adopt. Sourcesupport.zendesk.com/hc/en-us/articles/10488757995034read 2026-09-05 |
|
| Triggers & Channel Coverage | Full |
|
AI agents work across messaging, email, API, web form and voice channels, with each channel activated per agent, alongside the wider suite's social, chat and telephony intake. Sourcesupport.zendesk.com/hc/en-us/articles/8724978128282-getting-started-with-ai-agents-for-customer-serviceread 2026-09-05 |
|
| Model Flexibility & Routing | Partial |
|
Zendesk runs several third party model providers, using OpenAI, Google Gemini and ElevenLabs directly through zero retention endpoints and other models hosted on AWS Bedrock, Microsoft Azure, Groq and Fireworks AI. Zendesk makes the selection, and customers can only turn AI features on or off rather than choose the model that powers the product. Sourcesupport.zendesk.com/hc/en-us/articles/6059285322522-generative-ai-at-zendeskread 2026-09-05 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
REST APIs cover Support, AI Agents, Conversations, Talk, Chat, Custom Objects, Omnichannel, Search and Firehose. The Zendesk Apps framework with the ZAF client SDK builds installable apps, alongside mobile messaging SDKs for iOS and Android, a Web Widget API, a Channel framework and webhooks for real time events. The AI Agents API programmatically manages conversations, ticket workflows, escalations and data exports, gated behind the AI agents Advanced add on. Sourcedeveloper.zendesk.com/api-referenceread 2026-09-05 |
|
| Testing, Debugging & Optimization | Partial |
|
Zendesk QA is a shipped customer facing product that scores conversations produced by both human and AI agents, with AI generated prompt suggestions and prompt based spotlight insights. It grades what the agent has already produced, and there is no way to simulate, run scenarios against or regression test an agent before it goes live. The Resolution Learning Loop improves Zendesk's own models. Sourcesupport.zendesk.com/hc/en-us/articles/9224552305946-about-ai-insights-in-zendesk-qaread 2026-09-05 |
|
| Browser & Computer Use | Not documented |
|
AI agents act through authorized actions, API integrations and channel connectors, all programmatic surfaces. There is no browser control, virtual desktop or operation of software that lacks a programmatic interface. Sourcesupport.zendesk.com/hc/en-us/articles/10783053154202read 2026-09-05 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Product updates
Admins and agents with permission can now add internal notes, tags and field updates to a ticket while a Zendesk AI agent is still in the conversation, where these tickets were read only until it ended. They can also take over the ticket, which stops the AI agent, assigns the ticket to them and records the takeover in the ticket events, under two new custom role permissions.
Bears on: Observability / auditability
View sourceA new Ask for details step lets Zendesk AI agents collect structured details from customers mid conversation, filling ticket fields automatically before the flow continues. Rollout to all AI agent customers runs through October 6.
Bears on: Agent capability
View sourceZendesk made dedicated Helpjuice, WordPress, and ReadMe knowledge connectors generally available, with rollout beginning September 25. AI agents can use synced Helpjuice articles, public WordPress.com posts, and ReadMe guides. Private or self hosted WordPress content and ReadMe API references are excluded.
Bears on: Agent capability
View sourcePricing
Support Team at $19, Suite Team at $55 and Suite Professional at $115 per agent a month billed annually ($25, $69 and $149 billed monthly). Suite Enterprise + Copilot is sold through sales. AI agents are billed per automated resolution, and Copilot is $50 per agent a month.
Per agent seats, plus AI agent usage billed per automated resolution.
Included quota
Suite Team ($55) and Suite Professional ($115) include AI agents billed per automated resolution, with 5 automated resolutions per agent a month on Support Team and Suite Team and 10 on Suite Professional. Suite Professional includes the Data Center Location add on. Suite Enterprise + Copilot adds sandboxes, audit logs and approval workflows through sales.
What is public
Zendesk publishes per agent pricing billed yearly, with Support Team at $19, Suite Team at $55 and Suite Professional at $115, while Suite Enterprise + Copilot is sold through sales. AI agents are included in every Suite and Support plan and billed by automated resolutions. The Copilot add on is $50 per agent a month, and a 14 day free trial is offered.
Billing mechanics
Per agent seat subscription plus AI agent usage billed by automated resolution, a request the AI resolves without escalation. Add ons stack per agent, and enterprise plans generally go through sales.
Cost watchouts
A realistic stack adds Copilot and other per agent add ons to the seat price, and automated resolutions are billed on top.
Variable cost rationale
Two things are billed. Seats are a fixed price per agent a month, and add ons such as Copilot scale per agent too. AI agents are metered by automated resolution, a request the AI resolves without escalation, so that part of the bill grows with queue volume and with how often the AI succeeds, not with seats. Each plan includes a monthly allowance of automated resolutions per agent, 5 on Support Team and Suite Team and 10 on Suite Professional. Beyond it, resolutions cost $1.50 each when committed in advance or $2.00 pay as you go, so a prepaid commitment lowers the rate.
Additional watchouts
Add on stacking is the main cost driver. Per resolution AI billing rises with success, and the Enterprise tier is quoted.
Overage / add-ons
Automated resolutions beyond the included allowance cost $1.50 each when committed or $2.00 pay as you go.
Sales call required
Mixed (some tiers require a call)
Free / trial
14 day free trial with Suite Professional features and Copilot included by default
Lowest paid plan
Support Team at $19 per agent per month, billed annually ($25 monthly)
Commercial notes
Used by more than 100,000 companies, including Uber, Instacart, Zoom and Siemens.
Key ambiguities
The Enterprise tier is quoted. Automated resolutions carry a published rate, but how many a queue produces depends on how often the AI resolves requests without escalation, so that line of the bill is hard to predict.
Cancellation / refund
Month to month, annual or multiyear subscriptions, 20% off when billed annually, with volume terms negotiated. Plan changes apply immediately or at renewal, depending on the terms, and qualifying startups get a 6 month Suite trial for up to 50 agents.
Support SLA / resale
Support is tiered by plan, with Premier and Advanced support and Zendesk Assist sold as paid add ons. AI agents work in 80+ languages, Advanced AI agents build on the Ultimate.ai acquisition, and a large app marketplace is available.
Missing data
Suite Enterprise + Copilot prices are not published.
Related vendors
- Decagon — AI agent platform purpose-built for customer support, with deep…
- 5.Y — 5
- Ada — Agentic customer experience platform whose AI agents autonomously…
- Aisera — Enterprise agentic AI platform that orchestrates specialized agents…
- ASAPP — Generative AI for enterprise contact centers that plugs into…
- Assembled — AI customer support orchestration platform that unifies autonomous…
Alternatives to Zendesk
The closest documented capability profiles to Zendesk among customer support agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- NICE CXone10.5 / 14Matches Zendesk across all 14 documented capabilities
- BlueTweak9.5 / 14A lighter documented profile than Zendesk
- Chatbase10.0 / 14Fuller documented coverage on Model Flexibility & Routing
- Dialpad10.0 / 14Fuller documented coverage on Prebuilt Agents, Templates & Packs
- Fin11.0 / 14Fuller documented coverage on Prebuilt Agents, Templates & Packs and Testing, Debugging & OptimizationZendesk vs Fin →
- Freshworks10.0 / 14Fuller documented coverage on Prebuilt Agents, Templates & PacksZendesk vs Freshworks →
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded