Skip to main content
Zendesk chatbot integration

Custom AI chatbot integration for Zendesk

We build a custom AI chatbot and connect it to Zendesk through Sunshine Conversations. It answers in your messaging channels and passes conversations to your agents with the ticket already filled in.

Updated

Built on the switchboard

Your bot follows Zendesk's own control model, so agents receive conversations in Agent Workspace.

Tickets your team can use

Priority, group, tags and custom fields set at hand-off, with the conversation history attached.

Engineers from $45 an hour

Published rates, and a written quote after discovery for a defined project.

How does a custom AI chatbot connect to Zendesk?

Through Sunshine Conversations, Zendesk's messaging platform and API. You create a Conversations API key and a conversation integration with a webhook, so every customer message reaches your bot. The bot replies through the API, joins the switchboard that decides who holds each conversation, and passes control to Agent Workspace when a person is needed.

Zendesk describes Sunshine Conversations as the way to connect "another app or program, such as a custom bot" to your Zendesk instance, so agents answer every connected channel from Agent Workspace (Zendesk). The setup has five parts:

  • A Conversations API key, which authenticates the bot.
  • A conversation integration: a webhook and the events that trigger it.
  • The channel added to Agent Workspace.
  • A switchboard integration for the bot. Marketplace bots get one automatically; a custom bot is added through the Create Switchboard Integration API (Zendesk Developers).
  • A conversation control setting in Admin Center for resolved conversations: release control, or pass it on.

How any project runs is on our AI chatbot integration page. This page covers Zendesk.

How does the Zendesk switchboard pass control between the bot and an agent?

Only one integration controls a conversation at a time. Your bot can be the default responder, or come after Zendesk's own AI agent. When it needs a person it calls passControl, which makes Agent Workspace active at once and the ticket eligible for assignment. After releaseControl, a returning customer starts again with the default responder.

Each bot or agent system is a switchboard integration. In each conversation one is active and the rest wait on standby, without conversation events by default (Zendesk Developers). Four details shape the build:

  • Defaults per channel. The switchboard has a global default responder that each channel can override, so one channel can go straight to agents while web chat starts with the bot.
  • Use passControl, not offerControl. The protocol supports offering control, but Zendesk systems don't use that flow, so an offer to Agent Workspace is never accepted.
  • Use integration-scoped credentials. With them, passing control is idempotent; with app-scoped credentials, repeated calls can pass control more than once.
  • End with releaseControl. Passing control back to the default instead can leave a conversation stuck with an old default if you change it later.

For when a chatbot should hand off and what your agents should see, read what happens when the chatbot needs a person.

Our rates

Published prices in US dollars. Monthly plans are 160 hours, about 15% below the hourly rate.

Mid-level developer

3 to 5 years

$25 / hour

$3,400 a month (160 h)

Senior developer

6 or more years

$35 / hour

$4,750 a month (160 h)

AI / LLM engineer

Agents, RAG, IBM watsonx

$45 / hour

$6,100 a month (160 h)

Senior AI engineer

Leads AI architecture and delivery

$60 / hour

$8,150 a month (160 h)

Two-week trial with the developer you choose. No charge if you stop within the first five working days. US Eastern clients get 09:00 to 12:00 US Eastern time through an evening shift, at 15% extra. Full pricing and comparisons.

What can a custom bot read and write on Zendesk tickets?

While the bot is in control, Zendesk keeps the conversation as an AI agent ticket that agents can read but mostly can't edit, and triggers and SLAs don't run on it. At hand-off the bot can set priority, group, assignee, tags, brand, requester details and custom ticket fields, and Zendesk adds the recent conversation history.

Every conversation that doesn't start in Agent Workspace, including one with a third-party bot, creates an AI agent ticket. Until escalation it is read-only, except that admins and permitted agents can edit tags and custom fields and add internal notes (Zendesk). For a custom bot, that means:

  • Triggers, automations, SLAs, routing and macros wait until a person takes over. Messaging triggers still work.
  • Ticket list endpoints leave AI agent tickets out unless you add the support_type_scope=ai_agent parameter.
  • A bot-only ticket is solved after four days without a response, when the bot releases control, or when your bot's logic marks it resolved. It counts toward your storage limit.
Ticket data a custom bot can set when it passes control
Metadata keyWhat it sets
dataCapture.systemField.(field)Priority, organization_id, group_id, assignee_id, tags, brand_id, requester.name and requester.email (Zendesk Developers)
dataCapture.ticketField.(ID)Any custom ticket field, by its ID
first_message_idWhere the conversation history that Zendesk adds to the ticket begins
origin_source_typeThe ticket's channel, when you need to override Zendesk's detection

Custom field metadata counts only on the first escalation or when the bot releases control. The requester name is set only on the customer's first escalated conversation. Checked 5 November 2026.

Which Zendesk plan do you need for a custom chatbot?

Suite Professional, Enterprise or Enterprise Plus. Those plans include a low-volume, self-service version of Sunshine Conversations, with 1,000 monthly active users and 1,000 outbound notifications, so no separate license is needed. Busier accounts buy add-on packs. Plans bought before April 2023 are legacy plans with different limits.

What Zendesk includes and charges for Sunshine Conversations
ItemWhat Zendesk lists
Plans with accessSuite Professional, Enterprise and Enterprise Plus (Zendesk)
Monthly active users (MAU)1,000 included, then $50 for each pack of 2,500 more (Zendesk pricing)
Outbound notifications1,000 included, then $50 for each pack of 25,000 more
Suite Professional seatUS$115 per agent a month, paid yearly

As listed on Zendesk's pages on 5 November 2026. Confirm prices for your account before you budget.

An MAU is a unique end user who sends or receives a message or event with your integration in a calendar month. Proactive messages don't add MAUs, but one person on several devices or channels can count more than once. One note matters for a custom bot: when Sunshine Conversations connects a third-party endpoint, all MAUs are billable (Zendesk).

When are Zendesk's own AI agents the better buy?

When Zendesk is the only system involved, what Zendesk's AI agents do covers your needs, and your volume is modest. They are included in every Suite and Support plan, priced per automated resolution, and they already sit in the switchboard. A custom bot earns its cost when it needs your own model, data location or other systems.

Zendesk includes its AI agents in every Suite and Support plan and charges per automated resolution, a request the AI agent resolves without escalation. Its pricing page includes 5 resolutions per agent a month on Suite Team and 10 on Suite Professional, then US$2.00 each pay-as-you-go (Zendesk pricing). Its blog gives $1.50 per automated resolution, often defined by a 72-hour quiet period (Zendesk).

  • Choose Zendesk's agent when Zendesk is the only system involved, you have no rule on which model handles customer data, and nobody wants to own a bot.
  • Choose a custom bot when it must act in systems outside Zendesk, such as orders or billing, or the model must run in your country or on your servers (private AI).
  • Check volume. By our arithmetic, 5,000 automated resolutions a month at US$2.00 pay-as-you-go is $10,000 a month.

The two can work together. Zendesk documents a switchboard in which a custom integration first identifies the customer, then directs them to Zendesk's AI agent, another bot or Agent Workspace (Zendesk Developers). If Zendesk's own agent covers your need, we will tell you on the first call.

What does a Zendesk chatbot integration cost?

There are three parts: your Zendesk plan, Sunshine Conversations usage above the included 1,000 monthly active users, and the build. By our planning estimates, a bot that reads your business systems takes 480 to 960 hours, and one that takes actions takes 800 to 1,600. Engineering is $45 to $60 an hour.

Planning estimates for a first Zendesk chatbot
What the bot doesBuild hoursAt Webify.AI rates
Reads Zendesk and your business systems480 to 960$22,800 to $45,600
Also takes actions, such as updating an order800 to 1,600$42,000 to $84,000

Planning estimates, not quotes. Maintenance is 20 to 40 hours a month in the first months, $1,050 to $2,100. Zendesk fees and model usage come on top.

A first version scoped to one task with one or two integrations typically takes six to twelve weeks. See our AI agent cost guide and pricing page.

Frequently asked questions

Do we need a separate Sunshine Conversations license for a custom bot?

Not on Suite Professional, Enterprise or Enterprise Plus, which include a low-volume, self-service version. You buy add-on packs above 1,000 monthly active users or 1,000 notifications. Legacy plans bought before April 6, 2023 have their own limits.

Can our custom bot hand conversations to agents in Agent Workspace?

Yes. The bot calls passControl on the switchboard, which makes Agent Workspace active straight away and the ticket eligible for assignment.

Can the bot fill in ticket fields before an agent opens the ticket?

Yes. With passControl metadata it can set priority, group, assignee, tags, brand, requester details and custom fields. Custom fields are read only on the first escalation or when the bot releases control.

Why don't our triggers fire on bot conversations?

Zendesk keeps them as AI agent tickets until a person takes over. Ticket triggers, automations and SLAs start once the ticket is escalated. Messaging triggers work throughout.

Can Zendesk's AI agent and our own bot work in the same account?

Yes. Both are switchboard integrations, so a conversation can start with one and pass to the other, or to Agent Workspace, along your escalation path.

Get started

Talk to an engineer about your Zendesk chatbot

A conversation about your goals and systems, and a straight answer on whether an agent is the right tool. Paid work starts only after you have agreed a written scope.

  • We sign your NDA before the first call if you need one
  • You leave with a next step, even if the answer is "you don't need an agent"
  • The code and IP are yours if you go ahead

Book a free first call

Pick a time, or send us the details and we will reply.