---
title: "Zendesk Bot Builder & Essential AI EOL: The 2026 Migration Guide"
slug: zendesk-bot-builder-essential-ai-eol-the-2026-migration-guide
date: 2026-07-10
author: Rishabh
categories: [Zendesk]
excerpt: "Zendesk is shutting down bot builder and AI agents – Essential by December 10, 2026. Here's the timeline, rebuild work, pricing changes, and how to decide: stay or migrate."
tldr: "Zendesk bot builder shuts down December 10, 2026. No automatic migration exists — rebuild manually into the new AI agent system or use the forced rewrite to switch platforms."
canonical: https://clonepartner.com/blog/zendesk-bot-builder-essential-ai-eol-the-2026-migration-guide/
---

# Zendesk Bot Builder & Essential AI EOL: The 2026 Migration Guide


Zendesk is killing its legacy bot builder, AI agents – Essential, and default messaging response products. **Technical development stops August 31, 2026. Full service shut-off is December 10, 2026.** Every flow-based bot, Essential AI agent, and legacy autoreply configuration will stop working on that date.

There is no automatic migration path for existing bot builder flows. Zendesk has promised an automated in-product tool "later in 2026" with no confirmed date. The current guidance is a manual rebuild.

If your team still depends on legacy bots in production, you have roughly five months to either rebuild inside Zendesk's new agentic AI system or use this forced rewrite as the trigger to switch platforms entirely. This guide covers the exact timeline, what replaces what, the pricing impact, the manual rebuild work, and how to make the rebuild-vs-migrate decision.

## The Zendesk Bot Builder EOL Timeline

Here are the hard dates, confirmed in [Zendesk's official announcement](https://support.zendesk.com/hc/en-us/articles/10487730059034-Announcing-expanded-access-to-AI-agent-capabilities-for-all-Zendesk-customers):

| Date | What Happens |
| --- | --- |
| **May 11, 2026** | New AI agent packaging model begins phased rollout. Legacy products declared end-of-life. |
| **May 25 – June 12, 2026** | Existing Essential/basic customers get access to the new experience. |
| **August 31, 2026** | Zendesk stops all technical development for legacy products (critical bug fixes only). |
| **December 10, 2026** | Full service shut-off. Legacy products removed from the platform entirely. |

The affected products:

- **AI agents – Essential** (the generative AI tier included in Suite plans)
- **Bot builder** (flow-based answer builder with decision trees)
- **Bot builder answers and intents** (legacy since February 2, 2025)
- **Default messaging response** (auto-response greeting flows)
- **Autoreplies with articles** (legacy since July 31, 2025)

Bot builder, answers, and intents had already been treated as legacy since [February 2, 2025](https://support.zendesk.com/hc/en-us/articles/6970583409690-About-AI-agents). The 2026 announcements turned that earlier legacy status into a real shutdown clock. If you open an AI agent in Admin Center and still see an **Answers** tab or an **Intents** tab, Zendesk treats that configuration as legacy. That is the fastest way to identify accounts that need work.

> [!WARNING]
> **Treat August 31, 2026 as your internal migration deadline — not December 10.** After August 31, you are running on code that gets no updates. Any breaking change in a browser, channel, or integration could leave your bots dead with no fix coming. Zendesk explicitly recommends completing migration before the maintenance-mode cutoff.

## What Replaces Legacy Bots: The New AI Agent Architecture

The distinction between "Essential" and "Advanced" AI agent plans is gone. All customers transition to a single unified AI agent experience. Features previously restricted to the AI agents – Advanced add-on — agentic reasoning, multi-step procedures, external API integrations — are now included across Zendesk Suite and Support plans.

The architectural shift is significant. Legacy bot builder used **rigid decision trees** — if customer says X, respond with Y, branch to Z. The new system combines scripted dialogues with agentic reasoning, where AI dynamically works through a conversation, asks follow-up questions, and selects actions based on context.

A lot of teams talk about this change as if scripted bots are dead. That is not accurate. Zendesk's own [migration docs](https://support.zendesk.com/hc/en-us/articles/10543162665242-Migrating-to-the-new-AI-agents-experience) confirm that the new AI agents experience still supports scripted dialogues. The shift is that you can now choose where to keep explicit flow control and where to let a generative procedure handle ambiguity.

Here are the core replacement components:

### Knowledge Sources

Knowledge sources power AI-generated answers from your help center and external content. Zendesk lets one AI agent connect to multiple knowledge sources, but [warns that too many sources can reduce accuracy and increase latency](https://support.zendesk.com/hc/en-us/articles/8357749301658-Connecting-knowledge-sources-to-power-generative-replies-in-AI-agents).

### Use Cases

Use cases are the classifier layer. They decide what the customer is asking and route the request to the right dialogue or procedure. Use cases can connect to either dialogues or procedures, and messaging supports both, while email is procedure-only in the new agentic model.

### Dialogues

Dialogues are the scripted successor to legacy answer flows. Zendesk's own migration mapping says **Bot builder → Dialogue builder** and **Answers → Dialogues**. Use these when you need tight control over prompts, options, forms, or richer messaging behavior.

### Generative Procedures

Generative procedures are the agentic layer. You describe what the AI should do in natural language, and Zendesk generates a visual procedure map. The AI asks follow-up questions, chooses between available actions, and adapts within policy boundaries instead of following a rigid branch tree.

As of [June 11, 2026](https://support.zendesk.com/hc/en-us/articles/10904321914138-Release-notes-through-2026-06-12), you can edit individual blocks in the visual canvas instead of regenerating the entire procedure from scratch. You can update prompts, messages, conditions, actions, integrations, and switch block types. But there is a real constraint: **you cannot yet add, reorder, or delete blocks** when directly editing the procedure map. For complex flows, you are still heavily dependent on the AI regeneration step.

Procedures also have channel and formatting limits. **Rich-text formatting such as buttons and carousels is not supported in generative procedures.** For email, the AI cannot handle a message that mixes a knowledge question with a procedure request in the same reply — multi-procedure emails escalate to a human agent.

### Action Builder and External API Integrations

The biggest technical upgrade. AI agents can now call external systems via API to take real actions during conversations — look up an order, issue a refund, schedule a call. [Native connectors exist for HubSpot, Chargebee, Stripe, Calendly, and GitHub](https://support.zendesk.com/hc/en-us/articles/10943174976794-What-s-new-in-Zendesk-July-2026).

Zendesk restricts the integration builder to **Client Admins** and [warns that API descriptions must be explicit](https://support.zendesk.com/hc/en-us/articles/8357756844442-About-the-integration-builder-for-AI-agents) so the AI knows what each endpoint does, when to use it, and how parameters should behave. Poor descriptions directly affect runtime behavior — this is not a cosmetic issue.

### Cross-Channel AI Agents

One important constraint: in the new experience, **each AI agent can interact on only one channel type** (messaging, email, or voice). If you had one Essential AI agent deployed on both messaging and email, you now need [two separate AI agents](https://support.zendesk.com/hc/en-us/articles/10543162665242-Migrating-to-the-new-AI-agents-experience) to cover the same functionality. They share a unified knowledge base and procedure repository, but you are configuring and maintaining distinct agents per channel.

## The Pricing Model Shift

This is where the forced migration gets expensive. Legacy bot builder usage was effectively bundled into your seat price. The new AI agents bill on a **per-automated-resolution model** that stacks on top of existing per-seat costs.

**Zendesk does not publish list prices for automated resolutions on its public pricing pages as of July 2026.** The official docs explicitly state that invoice and service-order prices shown in documentation examples are [placeholders and direct customers to Sales for real pricing](https://support.zendesk.com/hc/en-us/articles/10479528943130-Upgrading-from-automated-resolutions-to-resolution-allowances). Commonly cited community and analyst figures put pricing at approximately **$2 per resolution on pay-as-you-go** and **$1.50 per resolution with committed volume**, but you should confirm against your actual contract or sales quote.

Each Suite plan includes a small free allotment — roughly 5 automated resolutions per agent per month on Team, 10 on Growth/Professional, and 15 on Enterprise, with a [stated maximum of 10,000 allocated automated resolutions per year](https://support.zendesk.com/hc/en-us/articles/6931689272090-Moving-to-automated-resolutions-from-existing-bot-pricing-plans) across plans.

### Resolution Tiers: What Counts Against Your Allowance

Not every AI interaction [consumes your allowance](https://support.zendesk.com/hc/en-us/articles/9570369117338-About-automated-resolution-tiers) the same way:

| Resolution Tier | Definition | Counts Against Allowance? |
| --- | --- | --- |
| **Assisted escalation** | The AI helped but a human agent resolved the issue. | **No** |
| **Contained resolution** | The AI handled the interaction and the customer did not ask for more help, but the post-conversation LLM check did not fully verify the outcome. | **No** |
| **Verified resolution** | After a **72-hour** window with no follow-up, Zendesk's verification process confirms the AI satisfactorily resolved the issue. | **Yes** |

Understanding the distinction matters for cost modeling. In practice, the ratio of verified resolutions to total AI interactions determines your effective cost per conversation. If 60% of AI interactions result in verified resolutions (a reasonable mid-range estimate for well-tuned bots with a solid knowledge base), your per-conversation cost is lower than the per-resolution rate — but you need to model with your own data. Pull your last 6 months of bot-handled conversations, estimate the verification rate based on your current resolution and escalation patterns, and apply your contract rate.

Zendesk's [forecast view](https://support.zendesk.com/hc/en-us/articles/8922391373978-Monitoring-automated-resolution-usage) uses 7 to 30 days of historical usage and a moving average model to estimate whether you will exhaust your allowance before the end of the cycle. Use this actively during your first billing cycle under the new model.

### Cross-Vendor Per-Resolution Pricing Comparison

If per-resolution pricing is a key factor in your rebuild-vs-migrate decision, normalize costs across vendors:

| Vendor | Per-Resolution Price | What Counts as a Resolution | Source |
| --- | --- | --- | --- |
| **Zendesk** | ~$1.50 (committed) / ~$2.00 (PAYG) | Verified resolution after 72-hour window, confirmed by LLM check | Contract/Sales (not publicly listed) |
| **Intercom (Fin)** | $0.99 per resolution | AI-resolved conversation where the customer's question was answered without human handoff | [Intercom pricing page](https://www.intercom.com/pricing) |
| **HubSpot (Breeze Customer Agent)** | $0.50 per resolved conversation | Resolved conversation as defined by [HubSpot's April 2026 announcement](https://www.hubspot.com/company-news/hubspots-customer-agent-and-prospecting-agent-now-you-pay-when-the-task-is-complete) | HubSpot company news |
| **Freshdesk (Freddy AI)** | Included in plan tiers; no separate per-resolution billing at comparable scale | N/A — bundled into seat pricing | Freshdesk pricing page |

**Caution:** These vendors define "resolution" differently. Zendesk's 72-hour verified window is stricter than Intercom's immediate resolution count. A direct dollar-to-dollar comparison requires understanding how each vendor counts. At 2,000 monthly resolutions, the annual cost difference between Zendesk ($36,000 at $1.50) and HubSpot ($12,000 at $0.50) is $24,000 — material enough to justify a serious evaluation.

### Budget Impact Example

For a 20-agent team on Suite Professional:

| Component | Monthly Cost |
| --- | --- |
| Suite Professional (20 agents × $115) | $2,300 |
| Free AR allowance | ~200 resolutions |
| Additional 2,000 resolutions × $1.50 | $3,000 |
| **Total monthly** | **$5,300+** |

That $3,000/month AI layer is a 130% increase over the seat-only cost. At 5,000 additional resolutions, the AI bill ($7,500) exceeds the seat cost ($2,300) by more than 3×.

> [!CAUTION]
> **Auto-billing is live.** Since January 2026, Zendesk auto-bills for resolution overages with no prior approval required. Before January 2026, overages above your committed monthly volume required manual activation. Budget accordingly — seasonal spikes can produce surprise invoices.

Before committing to the new system, model your resolution costs against actual ticket volume. If you are a high-volume support operation and the per-resolution math does not work, that changes the rebuild-vs-migrate calculus entirely.

## The Manual Rebuild Trap

The biggest operational problem is simple: **there is no automatic migration path for legacy bot builder flows today.** Zendesk has stated that an automated in-product migration tool will be offered ["later in 2026,"](https://support.zendesk.com/hc/en-us/articles/10543162665242-Migrating-to-the-new-AI-agents-experience) but no specific date has been confirmed. The same applies to default messaging responses.

"Later in 2026" with a December 10 kill date is not a comfortable timeline. If the tool ships in October, you get two months to test and deploy. If it ships in November, you are doing a rush migration during the holidays.

### Legacy-to-New Object Mapping

Zendesk publishes this mapping in its [migration docs](https://support.zendesk.com/hc/en-us/articles/10543162665242-Migrating-to-the-new-AI-agents-experience):

| Legacy Object | New Equivalent | What to Check |
| --- | --- | --- |
| Bot builder | Dialogue builder | Recreate scripted flow logic |
| Answers | Dialogues | Rebuild response blocks and control paths |
| Intents | Use cases | Re-train naming and description logic |
| Make API call | Integration or action flow | Re-test auth, payloads, and failures |
| Transfer to agent | Escalation | Preserve routing and queue behavior |
| Show help center articles | Generative replies / knowledge answers | Decide between curated vs. generated output |

### Dependency Chain: Triggers, Automations, and Views

Legacy bot objects are often referenced by other Zendesk configurations. **Triggers, automations, and views that reference legacy bot builder outcomes, intents, or answer IDs will break when those objects are removed.** Before rebuilding bots, audit your entire automation layer for dependencies:

- **Triggers** that route tickets based on bot builder intent matches or specific answer IDs
- **Automations** that escalate or tag tickets based on bot outcomes (e.g., "bot didn't resolve")
- **Views** filtered by bot-assigned tags or fields
- **Reporting dashboards** built on legacy bot metrics (containment rate, drop-off by step, intent match rate)

Export your trigger and automation rules and search for references to legacy bot objects. Any match needs to be remapped to the new AI agent's use cases, dialogue outcomes, or resolution tiers.

### Reporting and Analytics Migration

The metrics you lose and gain in the new system are different enough to affect dashboards and KPI tracking:

| Legacy Bot Metrics | New AI Agent Metrics |
| --- | --- |
| Containment rate (binary: resolved by bot or not) | Resolution tiers (assisted, contained, verified) |
| Drop-off rate by step | Conversation flow analysis across procedures |
| Intent match rate | Use case match confidence scores |
| Answer engagement (clicks, follows) | Knowledge source citation frequency |
| Session-level analytics | Resolution-level analytics with 72-hour verification window |

If you report bot containment rates to leadership, the shift to three-tier resolution tracking changes how you measure and report AI performance. Plan your dashboard migration alongside your bot migration.

### Sandbox Availability by Plan Tier

Testing a rebuilt automation layer in production is risky. Here is what Zendesk offers by plan:

| Zendesk Plan | Sandbox Access |
| --- | --- |
| Team | No sandbox |
| Growth | No sandbox |
| Professional | No sandbox (standard) |
| Enterprise | Premium sandbox included |
| Enterprise Plus | Premium sandbox included |

The Zendesk community is already feeling the pressure. One user noted: *"The schedule has barely 2 months from the end of the rollout window until the end-of-support date. We have a Professional instance — no sandbox. Any changes we need to make are to our production environment."* If you are on Team, Growth, or Professional, every change goes live immediately. Factor this constraint into your testing and rollout plan — consider deploying new AI agents to a low-traffic channel first as a de facto test environment.

### Why Manual Translation Fails

You cannot copy-paste legacy flow logic into a generative procedure.

1. **Intent mapping mismatch.** Legacy bots relied on strict keyword matching and predefined intents. Agentic AI uses semantic search. You need to audit legacy intents and rewrite them as natural language use case descriptions and boundary instructions.

2. **Knowledge base drift.** The new AI agents rely heavily on your Zendesk Guide content. If your knowledge base is outdated, poorly structured, or filled with contradictory articles, the agentic AI will produce confident but wrong replies. The legacy bot builder hid bad KB hygiene because you hardcoded the answers. The new system exposes it immediately.

3. **API payload recreation.** Any external system calls made by your legacy bots — whether through Make, Zapier, or custom webhooks — must be entirely rebuilt inside the new Action Builder or integration builder, with fresh authentication, header configuration, and JSON mapping.

4. **Channel duplication.** If your bots served multiple channels from a single agent, you now need separate AI agents for each channel type — doubling or tripling the configuration work.

5. **Step-level edge cases.** Legacy bot builder [supported up to 10 quick replies](https://support.zendesk.com/hc/en-us/articles/4408836323738-Understanding-the-step-types-for-AI-agent-answers-Legacy) in a Present options step, up to 10 panels in a carousel, up to 6 help center articles in an article step, and JSON responses under 2MB for Make API call. It also saved only the first 280 characters of a variable value in certain API-call patterns. If your legacy flows depended on those behaviors, validate each one explicitly during rebuild.

6. **Rollback constraints.** If the new AI agents perform worse than legacy bots in production, your rollback options narrow over time. Before August 31, you can revert to legacy bots while they are still maintained. After August 31, legacy bots still function but receive no bug fixes — any issue you encounter stays. After December 10, there is no rollback path. Test thoroughly before August 31 while you still have a safety net.

### Estimated Rebuild Time

Based on the scope of work involved — flow analysis, use case rewriting, knowledge base cleanup, API integration rebuilding, channel splitting, trigger remapping, and testing — here are rough time benchmarks for manual migration:

| Legacy Bot Complexity | Estimated Rebuild Time | Notes |
| --- | --- | --- |
| Simple (1–5 flows, FAQ-only, no API calls) | 8–20 hours | Mostly dialogue creation and KB cleanup |
| Moderate (6–20 flows, some API calls, 1–2 channels) | 40–80 hours | Includes integration rebuilds and channel splitting |
| Complex (20–50+ flows, multiple API integrations, 3+ channels, multiple brands) | 120–300+ hours | Full automation audit, trigger remapping, phased rollout required |

These estimates include testing time but assume a team already familiar with Zendesk Admin Center. Add 20–40% for teams new to the AI agent interface.

> [!TIP]
> **Do not attempt a 1:1 rebuild.** A legacy bot with 50 specific flows can often be consolidated into 5–10 generative procedures and a well-structured knowledge base. Use this decision framework: **dialogues** for tight flow control (order forms, data collection, branching menus), **generative procedures** for ambiguous multi-step requests (troubleshooting, policy questions with exceptions), and **knowledge-only AI answers** when the legacy flow was really just article retrieval wrapped in a bot shell. Treat this as a refactoring exercise, not a copy-paste job.

## Rebuild in Zendesk vs. Migrate Your Helpdesk Entirely

If you have to manually rebuild your entire automation layer anyway, is this the right time to evaluate whether Zendesk is still the right platform?

The bot builder EOL does not exist in a vacuum. Zendesk is pushing a massive amount of change onto customers simultaneously in 2026:

- **July 1 – September 30:** AI agents workspace migration and updates to knowledge and search rules
- **July 28:** [API token deactivation begins](https://support.zendesk.com/hc/en-us/sections/4405298833818). Unused tokens auto-deactivate. By April 30, 2027, all API tokens stop working — you must migrate integrations to OAuth.
- **August 31:** Bot builder and Essential AI development stops
- **October 28:** Remaining legacy CCs accounts forced onto CCs and followers
- **December 10:** Full legacy bot shut-off

Every custom integration using API tokens needs refactoring. Every bot needs rebuilding. Every automation needs reviewing. That is a lot of simultaneous engineering work for any operations team.

### Decision Framework: Rebuild vs. Migrate

Use this structured decision tree based on four variables — bot complexity, cost impact, concurrent change load, and platform requirements:

**→ If bot complexity is low** (fewer than 10 flows, mostly FAQ deflection) **AND** per-resolution pricing adds less than 30% to your current support costs **AND** you are already on Suite Professional or Enterprise: **Rebuild inside Zendesk.** The native agentic capabilities tied to Zendesk routing and reporting will outperform what you had.

**→ If bot complexity is high** (20+ flows with API integrations) **OR** per-resolution pricing increases costs by more than 50% **OR** you are simultaneously facing API token migration, AI workspace changes, and other forced changes: **Seriously evaluate migration.** The cumulative rebuild cost may exceed a full platform switch.

**→ If you need model choice, custom AI observability, or multi-vendor AI orchestration** that Zendesk's closed system does not offer: **Migration or an overlay architecture is the right path** regardless of bot complexity.

### The Over-the-Top AI Alternative

If you want to keep Zendesk Support for human agents but refuse the per-resolution pricing, some teams are deploying third-party AI layers that sit on top of Zendesk. Platforms like Voiceflow, Ada, and Cognigy position themselves around model choice, cross-channel deployment, and native Zendesk connectors that can create and update tickets, add comments, and answer from your Zendesk Help Center. These platforms intercept tickets, resolve them, and sync data back without triggering Zendesk's native automated resolution fee. The trade-off: you are now maintaining two systems, and Zendesk's roadmap may introduce restrictions on third-party bot access in future releases.

### The Full Helpdesk Migration Alternative

If the total cost of ownership has become unjustifiable, these platforms are worth serious comparison:

- **Intercom** — AI-first with Fin for automated resolution at $0.99 per outcome. Tighter product integration and a messaging-first architecture. See our [Zendesk to Intercom Migration Guide](https://clonepartner.com/blog/blog/zendesk-to-intercom-migration-the-2026-technical-guide/).
- **Freshdesk** — Bundled AI in plan tiers without separate per-resolution billing at Zendesk's scale. A cost-driven switch for teams who want to escape metered AI billing. Our [Zendesk to Freshdesk Migration Guide](https://clonepartner.com/blog/blog/zendesk-to-freshdesk-migration-2026-how-to-avoid-data-loss/) covers the data loss risks.
- **HubSpot Service Hub** — Strong fit if you are already in the HubSpot ecosystem. [Breeze Customer Agent moved to $0.50 per resolved conversation](https://www.hubspot.com/company-news/hubspots-customer-agent-and-prospecting-agent-now-you-pay-when-the-task-is-complete) in April 2026.

If you are evaluating alternatives, normalize every vendor quote to the same unit: **resolved conversations**, **verified outcomes**, or **sessions**. Comparing seat price alone will lead you into a bad budget model. Refer to the cross-vendor pricing table above for a starting framework.

## A Practical Migration Checklist

Whether you are staying on Zendesk or leaving, start these steps **now** — not in October.

1. **Inventory every legacy AI asset.** Pull a list of AI agents, answers, intents, default messaging responses, channels, brands, and API-backed flows. If you still see legacy Answers or Intents tabs, those assets need attention. Export the logic via screenshots and documentation before August 31 — Zendesk has not confirmed that legacy bot configurations will be accessible after the shut-off.

2. **Audit triggers, automations, and views for legacy bot dependencies.** Search your trigger and automation rules for references to legacy bot objects (intent IDs, answer IDs, bot tags). Document every dependency — these must be remapped to new AI agent objects during migration.

3. **Classify each flow before rebuilding.** Put every legacy flow into one of four buckets: retire, rebuild as dialogue, rebuild as generative procedure, or replace with knowledge-only AI answers.

4. **Split your channel plan early.** The new experience is channel-specific, so messaging and email coverage may need separate AI agents even when the old setup looked unified.

5. **Clean up your knowledge base.** Because the new AI relies on semantic search rather than hardcoded logic, update, consolidate, and format your Zendesk Guide articles. Reduce source sprawl — too many knowledge sources hurts accuracy and latency.

6. **Rebuild API integrations deliberately.** Confirm the right admins own the integrations and that descriptions are explicit enough for the AI to act correctly. Re-test authentication, payloads, and failure handling in the new Action Builder.

7. **Audit your API token usage.** With the July 28 auto-deactivation coming, identify every integration using API tokens and plan the OAuth migration in parallel with your bot rebuild.

8. **Model your resolution costs.** Pull your last 6 months of bot-handled conversations. Estimate your verified-resolution rate based on current escalation and resolution patterns. Apply the per-resolution rate from your contract. Compare against current costs. Use Zendesk's resolution usage and forecast tooling to set a contract-aware buffer before production traffic moves.

9. **Test using Zendesk's AI agent testing tools.** Use the [conversation testing mode](https://support.zendesk.com/hc/en-us/articles/8357749301658-Connecting-knowledge-sources-to-power-generative-replies-in-AI-agents) to simulate interactions before publishing. Stand up a test AI agent, connect your knowledge base, and build generative procedures to understand how the new system handles your use cases. Review conversation logs and the verification details panel for each resolution. Compare new behavior against live historical cases, not just happy-path demos. If you lack sandbox access (Team/Growth/Professional), deploy to a low-traffic channel first.

10. **Plan your reporting migration.** Map your current bot performance dashboards to the new metrics (resolution tiers, confidence scores, knowledge citation rates). Build new dashboards before cutover so you can compare performance between legacy and new systems during the transition.

11. **Set your decision deadline.** If you are going to leave Zendesk, decide by mid-August. A full helpdesk migration takes 1–4 weeks depending on data volume and complexity. Waiting until November is a gamble. If you are staying, aim to complete migration before August 31 to avoid running in maintenance-mode territory while carrying production risk.

If there is any chance you are leaving, start your [data export](https://clonepartner.com/blog/blog/how-to-export-tickets-from-zendesk/) now. Zendesk's native export has limits — you may need API-based extraction for full history, attachments, and CSAT scores.

## When to Bring in Migration Specialists

Some teams can handle a 5-flow rebuild internally. But for organizations running dozens of bot builder flows across multiple brands and channels, with hundreds of thousands of historical tickets, the forced-rebuild timeline creates real operational risk.

The work is usually not "export the tickets and hope." It is inventorying brittle legacy logic, auditing trigger and automation dependencies, mapping flows to dialogues or procedures, extracting historical tickets and knowledge, handling attachments and inline images that standard export tools drop, running a parallel validation pass, and cutting over without breaking support operations.

If the bot EOL is just one item in a larger Zendesk change backlog, bringing in an engineering-led migration partner is often cheaper than tying up your product and support ops teams for a quarter on vendor-driven rework.

At ClonePartner, we handle both scenarios — staying on Zendesk with a rebuilt automation layer, or migrating your entire helpdesk to a new platform:

1. **Deep extraction.** We pull everything — not just standard fields, but custom objects, side conversations, internal notes, and inline images that standard export tools drop.
2. **Logic mapping.** We analyze your legacy bot flows, macros, triggers, and automations, and map that business logic to your target platform's automation engine.
3. **Shadow mode execution.** We run parallel syncs to ensure your new helpdesk is routing tickets correctly before you change your MX records or web widgets.
4. **Delta sync.** We perform a final delta sync on go-live day, ensuring tickets created during the migration window are captured perfectly.

The December 10 deadline is fixed. The question is whether you spend the next five months rebuilding internally or bring in a team that has done this hundreds of times.

> Facing the Zendesk bot builder EOL and not sure whether to rebuild or migrate? Book a free 30-minute call with our engineering team. We'll help you assess your options, scope the work, and build a plan that hits the December 10 deadline — with room to spare.
>
> [Talk to us](https://cal.com/clonepartner/meet?duration=30)

## Frequently asked questions

### When does Zendesk bot builder stop working?

Zendesk stops technical development for the legacy bot builder on August 31, 2026, with critical bug fixes only after that date. The full service shut-off — where legacy bots are removed from the product entirely and stop functioning — is December 10, 2026.

### Is there an automatic migration tool for Zendesk bot builder flows?

Not yet. Zendesk has promised an automated in-product migration tool 'later in 2026,' but no specific date has been confirmed. Currently, you must manually rebuild your bot builder flows as dialogues or generative procedures in the new AI agents experience.

### How much does Zendesk charge per automated resolution?

Commonly cited figures are approximately $1.50 per automated resolution with committed volume and $2.00 on pay-as-you-go. Each Suite plan includes a small free allowance (roughly 5–15 resolutions per agent per month). However, Zendesk's newer resolution-tier documentation uses placeholder prices and directs customers to Sales for actual contract pricing. Since January 2026, overages are auto-billed with no prior approval required.

### What replaces Zendesk AI agents Essential?

Zendesk replaced the Essential/Advanced distinction with a single unified AI agent experience available to all Suite and Support plans. It includes knowledge sources, use cases, scripted dialogues (the successor to bot builder flows), generative procedures with agentic reasoning, external API integrations via Action Builder, and cross-channel AI agents for messaging, email, and voice.

### Should I rebuild Zendesk bots or migrate to a different helpdesk?

Rebuild inside Zendesk if you have fewer than 10 simple flows, per-resolution pricing works for your volume, and your team has bandwidth before August 31. Consider migrating if you have dozens of complex flows, per-resolution costs are prohibitive, and you're also facing API token deprecation and other forced changes in the same window. Since you're effectively doing implementation-level work either way, the forced rebuild is a natural trigger to evaluate alternatives.