How to Set Up Zendesk AI Agents (Answer Bot) — Step by Step

This guide is the current, verified version. We'll map the old name to the new one, then walk through setting up a Zendesk AI agent step by step — enabling it, building a conversation flow, connecting it to your messaging and email channels, testing it, and measuring whether it actually resolves anything. Everything below is checked against Zendesk's own documentation as of June 2026, because this is a topic where most of what's online is out of date.

Answer Bot vs. AI agents: the naming, untangled

A 60-second history, because the names genuinely matter when you're clicking through the admin:

• Answer Bot (2017–2024) was Zendesk's article-suggestion bot. It matched a customer's question to your help-center content and replied with relevant articles. Useful, but it deflected — it didn't converse.
• AI agents is the current umbrella term. In 2024 Zendesk consolidated Answer Bot and its rule-based bot builder into the "AI agents" branding, and the old rule-based "bot builder, answers, and intents" experience (what many people called Flow Builder) was marked legacy on February 2, 2025.
• Today there are two flavors, and the distinction is the single most important thing to get right before you start:

One critical 2026 caveat. Zendesk is in the middle of merging these tiers. In a March 2026 announcement, "Announcing expanded access to AI agent capabilities for all Zendesk customers," Zendesk said it is removing the Essential/Advanced split and bundling Advanced capabilities (agentic reasoning, multi-step procedures, external API integrations) into Suite and Support plans. The rollout runs from May 11, 2026 for new accounts through mid-June 2026 for existing ones. At the same time, AI agents — Essential is now legacy (development and bug fixes stop August 31, 2026; full removal is scheduled for the end of 2026). Net effect: if you're setting up today, you're building on the newer AI-agent experience, not the old Essential bot — but the underlying steps below are the same.

For the bigger picture of what Zendesk's AI does across the product, see our explainer on Zendesk AI.

Before you start: prerequisites

A few things save you a mid-setup detour:

• Admin access. AI agents are configured in Admin Center, which needs an admin role.
• A populated help center. AI agents — Essential answers primarily from your Help Center articles. If your knowledge base is thin or written in internal jargon, the bot has nothing good to say. Clean, customer-language articles are the single biggest lever on bot quality.
• A channel to put it on. A messaging channel (Web Widget or mobile SDK) for conversational bots, and/or email/web form for autoreplies. If you haven't turned messaging on yet, do that first — our Zendesk messaging setup guide covers it, and it matters here because turning on messaging auto-creates a default AI agent for you.
• A supported plan. AI agent functionality is available across Zendesk Suite (Team, Growth, Professional, Enterprise, Enterprise Plus) and Support (Team, Professional, Enterprise).

Step 1 — Find (or create) your AI agent

There are two ways an AI agent comes into existence in Zendesk, and knowing which applies to you avoids a "where is it?" moment.

If you've already turned on messaging: you have one already. Zendesk automatically creates a default AI agent for each messaging channel when messaging is enabled — it greets customers and answers from your help center out of the box. You don't create it; you go configure it.

To find or create one manually:

1. Open Admin Center.
2. In the sidebar, click Channels.
3. Select AI agents (in some accounts this section reads AI agents and automation).
4. You'll see your existing AI agents listed. To build a new one, click Create AI agent (or the equivalent) and choose the channel type it serves.

The path to remember is Admin Center → Channels → AI agents. This is the home base for everything that follows — the conversation builder, the channel connections, and the autoreply settings all hang off here.

Step 2 — Build the conversation flow

This is where an AI agent stops being a generic text box and becomes useful. For a messaging AI agent, you build the experience in the bot builder (the flow editor). The pieces you'll configure:

1. The greeting. The first message a customer sees when they open the widget. Make it specific — "Hi! Ask me about orders, returns, or your account" beats "How can I help?" because it primes better questions and signals what the bot can actually do.
2. Answers from your help center. The core of an Essential bot. You define answers — responses triggered when a customer's message matches a topic — and let the bot surface the relevant Help Center articles. This is the modern form of what Answer Bot always did, now inside a conversational flow.
3. Custom flows. Beyond article suggestions, you can build step-by-step branches: ask a clarifying question, present options as buttons, collect details (order number, email), and route based on the answer. This is how you handle structured, repeatable requests ("track my order," "reset my password") end to end.
4. Business hours. Branch the conversation on whether your team is online. Off-hours, set the right expectation ("We're away until 9am — leave your email and we'll follow up") instead of implying a human is standing by.
5. Handoff to a human. The most important branch. Define when the bot gives up — a "talk to a person" option, or an automatic escalation when it can't match an answer — so the conversation transfers to an agent and lands as a ticket with the full transcript attached. A bot that can't gracefully hand off is worse than no bot.

A note on the old rule-based builder: the legacy "answers and intents" Flow Builder still functions during the migration window, but Zendesk is steering setups toward the newer generative AI-agent experience, which blends scripted flows with generated replies. Build new flows in the current builder rather than investing in the deprecated one.

Step 3 — Connect it to your channels

An AI agent only does something once it's attached to a channel. There are two distinct surfaces, and they behave differently.

Messaging (Web Widget and mobile SDK). This is the conversational bot. If the agent was auto-created with messaging, it's already connected to that channel — you just refine it. If you built one manually, attach it to the right messaging channel and brand. For multi-brand accounts, each brand can run its own AI agent with its own greeting and flows. (Setting up the underlying Web Widget/mobile channels themselves is covered in the messaging setup guide.)

Email and web form autoreplies. This is the part people forget. AI agents can also send autoreplies to incoming email and web form tickets — automatically replying with suggested help-center articles before a human picks it up. This is the most direct descendant of classic Answer Bot, and it's configured separately from the messaging flow (through your autoreply/trigger setup rather than the conversation builder). If a chunk of your volume still arrives by email, turning on autoreplies is low-effort deflection.

The payoff of connecting both: the same knowledge base powers your widget conversations and your email autoreplies, so customers get consistent answers regardless of how they reached you.

Step 4 — Test before you ship

Never announce a bot you haven't broken yourself. Run the full loop:

1. Preview in the builder. Most flows can be previewed inside the bot builder — step through the greeting and a couple of branches to confirm the logic flows.
2. Test on the live channel. Load the Web Widget on a staging or live page, open it, and ask a question your bot should handle. Confirm it answers and surfaces the right article.
3. Force a handoff. Ask something deliberately out of scope. Verify the bot escalates, the conversation becomes a ticket in the Agent Workspace, and it's routed to the right group with the transcript attached. (If you're fuzzy on how tickets are created and routed, see how the Zendesk ticketing system works.)
4. Check off-hours behavior. If you set business-hours branching, test it outside your hours so the away message actually fires.

Ninety percent of "the bot isn't working" reports trace back to three things: the widget snippet isn't on the page, the help center the bot draws from is empty, or the handoff branch was never configured.

Step 5 — Measure what it actually resolves

A live bot isn't a finished bot. Zendesk reports AI-agent performance so you can tell deflection theater from real resolution.

• Automated resolutions. This is the headline metric — and, importantly, the billing unit. Zendesk replaced the old per-bot / monthly-active-user "Answer Bot resolutions" model with automated resolutions: a count of issues the AI agent resolved without a human. (Per Zendesk's "Moving to automated resolutions" guidance, unused Answer Bot resolutions or MAUs convert into automated resolutions.)
• CSAT. Track satisfaction on bot-handled conversations specifically. A bot can "resolve" a ticket by exhausting the customer's patience; CSAT is the honesty check.
• Handoff and escalation rate. How often the bot punts. Rising escalations usually mean a knowledge gap you can close with a new article.

The billing nuance worth understanding

This is where teams get surprised. Zendesk bills AI agents per automated resolution, with an included monthly allowance that scales by plan and agent count — roughly 5 resolutions per agent on Team, 10 on Growth/Professional, and 15 on Enterprise each month, after which overage applies. (Allowances per Zendesk docs; verify against your contract.)

A few third-party observations to flag (these are estimates from analysts like eesel AI, not official Zendesk list prices — treat as approximate and confirm): the per-automated-resolution overage rate is commonly estimated around $1.20–$1.50, and not every AI interaction counts as billable — some interpretations distinguish a "verified resolution" (billable) from an escalation or an unconfirmed reply (not). The practical takeaway holds regardless of the exact figure: your cost scales with the number of issues the bot closes, so at high volume the math compounds. Model it against your conversation volume before you commit.

Where Macha fits in (honestly)

Now the direct part, since you're here for AI specifically. Zendesk's built-in AI agent is genuinely good at two things: deflecting with help-center article suggestions, and running known, scripted flows ("track my order," "reset password"). Where it tends to stall is the long tail — the questions whose answer lives in a past ticket, a policy doc, or another system, and that don't map to a tidy pre-built flow. Those still escalate.

That's the gap Macha is built for, and it's worth being precise about what Macha is and isn't. Macha is not a help desk and not a Zendesk replacement. It's an AI agent layer that runs on top of your existing Zendesk — another integration you connect, plainly. What it adds beyond the built-in bot: instead of only suggesting an article or escalating, it reads the actual question, pulls from your connected knowledge and past tickets across channels, and resolves the issue inside the same thread — then does the ticket housekeeping (tagging, status, routing) and hands off to a human, with context attached, when it isn't confident.

There's a real pricing difference, too, and it's not a gotcha — it's a genuinely different model. Zendesk bills per automated resolution. Macha bills per AI action — every automated step it takes (drafting a reply, tagging, routing, looking something up, resolving) — not per closed conversation. The reasoning: most automation isn't one clean "resolution," it's a series of actions done along the way, so charging per action reflects the actual work rather than betting everything on an outcome label. Whether that's cheaper for you depends entirely on your volume and how much per-ticket work you want automated — so model both.

The honest caveat: Macha is only as good as the knowledge you connect, and it's one more thing to configure and maintain. If your volume is mostly repetitive questions your help center could answer, the built-in Essential bot may be plenty. If you keep hitting its ceiling, that's the line where a layer like Macha earns its place. You can try it free — 7-day free trial, no credit card required.

Best practices and common mistakes

• Fix your help center first. A bot grounded in thin or jargon-heavy content will confidently give wrong answers. Knowledge quality is the ceiling on bot quality.
• Configure the handoff before the greeting. Everyone perfects the welcome message; the escalation path is where setups actually fail. Test it first.
• Turn on email autoreplies, not just messaging. If email is a big channel for you, AI-agent autoreplies are free deflection you've probably left switched off.
• **Watch automated resolutions and CSAT together.** A high resolution count with sinking CSAT means the bot is closing tickets by frustrating people, not helping them.
• Don't build on the deprecated Flow Builder. With the legacy "answers and intents" experience sunsetting at the end of 2026, invest your flow-building time in the current AI-agent builder.
• Model the billing before you scale. Per-automated-resolution pricing is cheap at low volume and material at high volume. Know your number.

Frequently asked questions

Is Zendesk Answer Bot still a thing? Not under that name. Answer Bot was Zendesk's 2017 article-suggestion feature; it was folded into AI agents in 2024, and the old rule-based bot builder became legacy on February 2, 2025. The capability lives on as AI agents — Essential (and its email/web-form autoreplies), so if a tutorial tells you to enable "Answer Bot," look under Admin Center → Channels → AI agents instead.

Where do I set up a Zendesk AI agent? Admin Center → Channels → AI agents (sometimes labeled "AI agents and automation"). If you've already turned on messaging, a default AI agent was created automatically — you configure that one rather than starting from scratch.

What's the difference between AI agents — Essential and Advanced? Essential is the included bot grounded in your help center (the Answer Bot/Flow Builder descendant). Advanced is the more autonomous, Ultimate-powered option with dialogues, actions, and API integrations. Note that as of 2026 Zendesk is removing the split and bundling Advanced capabilities into Suite and Support plans, while Essential becomes legacy.

Is the Zendesk bot free? AI agent functionality is included across Suite and Support plans, but usage is billed by automated resolution above a monthly allowance (roughly 5/agent on Team, 10 on Growth/Professional, 15 on Enterprise). So it's "included" up to a point, then metered. Confirm specifics against your plan.

How is Zendesk's AI agent billed? Per automated resolution — a count of issues resolved without a human — which replaced the older per-bot/MAU model. Each plan includes an allowance; overage is billed beyond it. (Per-resolution dollar rates circulating online are third-party estimates, not official Zendesk list prices.)

Can the AI agent answer email, not just chat? Yes. Separate from the messaging flow, AI agents can send autoreplies to email and web form tickets, replying with suggested articles before an agent steps in. It's the most direct continuation of classic Answer Bot.

The bottom line

Setting up a Zendesk "Answer Bot" today means setting up an AI agent: confirm whether you already have a default one (you do, if messaging is on), find it under Admin Center → Channels → AI agents, build the conversation flow (greeting, help-center answers, custom flows, business hours, and a clean handoff), connect it to messaging and email autoreplies, test the full loop — especially the escalation — then measure automated resolutions and CSAT while keeping an eye on the per-resolution billing. Get the naming straight, ground it in a solid help center, and the built-in bot will handle a real share of your volume. When you hit the wall on the questions it can only deflect or escalate, that's where an on-top layer like Macha extends what Zendesk's bot starts.

Verified against Zendesk's official documentation, June 2026. Zendesk is actively migrating its AI-agent tiers through 2026 — confirm labels, plan inclusions, and pricing in your own account before relying on them.