Most of your queue asks the same twenty questions. A person should not type those answers a thousand times — Zendesk AI agents resolve customer requests on messaging, email and voice, billed per automated resolution — and through Forethought, they also work on help desks that are not Zendesk.
Buy through TechBag
Same software. Better outcome — at a lower cost.
How it’s rated
Full scoreboard ↓Quick answer
This page covers Zendesk AI Agents — including Forethought AI Agents by Zendesk and the advanced AI agents. The rest:
Most product pages skip this. We start here — so you buy a capability, not a buzzword.
An AI agent reads a customer’s request and resolves it on its own, answering or acting, and hands over to a person when it cannot.
What consolidation actually replaces, dimension by dimension.
| Dimension | Scripted chatbot and seat-priced AI | Zendesk AI Agents |
|---|---|---|
| Repeat questions | Each one waits in the queue for a person | An AI agent answers at any hour |
| How AI is charged | Per seat, used or not | Per automated resolution, after an allowance |
| Keeping your help desk | Migrate first, then get the AI | Forethought agents sit on the desk you have |
| Changing what the bot does | A developer edits decision trees | A support lead edits it in Agent Builder |
| Checking AI answers | Spot-checks after a complaint | QA scores AI chats beside human ones |
| What it is NOT | — | India-hosted, INR-billed, or a sure 80% |
The cheapest test is the 14-day Suite Professional trial: point one AI agent at your top intents and count what it truly resolves.
Vendors love diagrams; buyers need to know what they’re actually operating. Here’s the whole platform, demystified.
Agents reply in the messaging widget, over email and on calls, drawing on your knowledge base, and pass the conversation to a human agent when they cannot resolve it themselves.
Zendesk calls the agents self-improving: its Resolution Learning Loop feeds outcomes back into them, and the company says its AI draws on roughly 20 billion ticket interactions.
Agent Builder is no-code, so a support lead rather than a developer defines each Custom Agent; Industry Agents ship ready-made, starting with commerce and then other sectors.
Forethought AI Agents by Zendesk work on any platform without a migration, the advanced agents connect to Freshdesk, Intercom and Freshchat, and agents also run in Salesforce and ServiceNow.
Agents on messaging, email and voice — billed per resolution, and able to run on Zendesk or on another help desk.
Zendesk AI agents resolve customer requests on their own, and Zendesk bills for each automated resolution.
Agents reply inside the messaging and live chat that Suite plans include, and hand over to a person with the conversation attached.
The same agents take the email channel, so a returns or order-status question that arrives by mail can be closed by the AI.
Voice AI agents take calls; a January 2025 spotlight names PolyAI in its title, with call resolution of up to 80% as the vendor’s pitch.
Support leads set up Custom Agents without code; Zendesk counts over 1 million Custom Agent executions in seven weeks.
Prebuilt Industry Agents launched for commerce, with financial services, media and technology versions promised next.
Suite Team and above include Action builder, so an agent can call another system for an order look-up instead of quoting an article.
Billing counts automated resolutions, not seats: 5, 10 or 15 per agent a month come with the tier, then US$2.00 each.
Zendesk QA, a US$35 add-on, scores AI-agent conversations beside human ones, so a weak answer surfaces before customers complain.
A location add-on pins core data to the US, EEA, UK, Japan or Australia, but the advanced agents run on Google Cloud outside it.
Zendesk’s 2025 launch of its AI agents, the March 2026 Forethought deal, a 2026 Custom Agents demo and a January 2025 voice spotlight, all from the official Zendesk channel.
Zendesk’s launch film for its AI agents, setting out the resolution-first pitch behind the per-resolution price.
The Forethought deal explained: self-improving agents that can also sit on a help desk that is not Zendesk.
A walk-through of setting up a Custom Agent in Zendesk without writing code.
Voice AI agents as Zendesk showed them in January 2025, with PolyAI in the title; confirm what ships today.
Want a live, India-context walkthrough for your environment?
Book a guided demo →Here’s what genuinely sets it apart — and exactly where it stops.
Zendesk charges for its AI agents by automated resolution rather than by seat or by conversation. Every Suite and Support plan carries 5, 10 or 15 resolutions per agent a month depending on the tier, capped at 10,000 a year, and pay-as-you-go resolutions beyond that cost US$2.00 each.
Zendesk closed the Forethought deal on 26 March 2026, and Forethought AI Agents by Zendesk work on any platform. The advanced agents from the Ultimate acquisition connect to Freshdesk, Intercom and Freshchat, and Zendesk says its agents run in Salesforce and ServiceNow too.
Inside Zendesk the agents use the knowledge base, messaging and Action builder that Suite plans include, and Zendesk QA can grade their conversations next to human ones. Agent Builder needs no code, and Zendesk reports more than 1 million Custom Agent executions in the seven weeks after launch; SodaStream and Tonies are quoted.
Indian buyers get no local hosting: the five locations on offer are American, European, British, Japanese and Australian, the add-on that pins one must be turned on, and the advanced agents run on Google Cloud beyond its reach. Nothing is invoiced in rupees. Forethought on its own is quote-only, and the 80% figure is Zendesk’s claim, not a promise.
Export three months of tickets, group the twenty most common requests, and mark which need an action in another system.
Estimate monthly automated resolutions, price the excess over the allowance at US$2.00, and set that beside rival rates.
In Agent Builder, make one answer-only agent and one that calls your order system through an action; test both on staff.
Switch the agent on in messaging first, grade its chats in QA each week, and fix the article behind every wrong answer.
Activate the data-location add-on, record that advanced agents run on Google Cloud, then add email and voice if results hold.
Modelled on Gartner Peer Insights structure. *Counts and breakdowns are illustrative pending verified review collection.
“Order-status chats were a third of our queue. The AI agent now closes most of them, and we budget the bill in resolutions.”
“We kept Freshdesk and put Forethought agents on top of it. No ticket migration, and that is what won over our CIO.”
“Plan for overage. In our festive-sale month the included resolutions ran out in week three and pay-as-you-go took over.”
“We graded the AI’s chats in QA beside the team’s. It caught two help articles still quoting last year’s refund rules.”
“Our DPO wanted AI conversations hosted in India. Zendesk has no region here, so we chose the EEA and wrote down why.”
“Agent Builder let our support lead set up the warranty-claim flow herself; IT only had to connect the order system.”
Analyst firms bury this view behind paywalls, and G2 retired its Grid. So here’s TechBag’s synthesis of the AI customer service market — tap any vendor to see why it sits where it does.
Execution strength vs product vision — the classic market map, minus the paywall.
US$2.00 per resolution after a per-agent plan allowance.
The grid nobody publishes — how clearly a vendor offers storage in India vs how far its AI agent reaches beyond its own help desk.
No India region; Forethought runs on any help desk.
Positions are TechBag’s illustrative synthesis of public review-platform data and vendor documentation — not a reproduction of any analyst graphic. Verify before relying on it.
Set beside Salesforce Agentforce, Intercom Fin, Freshworks Freddy AI Agent, HubSpot Breeze Customer Agent and ServiceNow AI Agents — on price, reach, building, oversight, voice and India.
| Dimension | Zendesk AI Agents | Salesforce Agentforce | Intercom Fin | Freshworks Freddy AI Agent | HubSpot Breeze Customer Agent | ServiceNow AI Agents |
|---|---|---|---|---|---|---|
| What it is | Resolution-billed agents | Agentic AI platform | Standalone AI agent | Freshdesk’s AI agent | HubSpot’s service agent | Agents on Now Platform |
| Deployment | SaaS on AWS and GCP | SaaS on Hyperforce | Hosted, no migration | Freshworks SaaS | Inside HubSpot | On a Now instance |
| Channels | Messaging, email, voice | Service, Slack, voice | Chat, email, voice, SMS | Chat apps and email | Chat, email, WhatsApp | Portal, Virtual Agent |
| Other help desks | Yes, via Forethought | Salesforce-centred | Many desks | Part of Freshdesk | Part of HubSpot | ServiceNow instance |
| Pricing model | Per resolution | Credits, chats or users | Per outcome | Per session | Per resolved chat | Not published |
| Published entry price | US$2.00 pay-as-you-go | US$2 a conversation | US$0.99 an outcome | US$49 per 100 sessions | US$0.50 a resolution | Quote only |
| Included vs add-on | Allowance in every plan | Humans need licences | Every outcome is billed | Copilot costs extra | Pro or Enterprise only | AI in every package |
| Resolution claim | Up to 80% (claim) | No rate published | 76% average (claim) | Not published | 65% (claim) | Not published |
| Building agents | No-code Agent Builder | Plain-language builder | Procedures, no engineers | No-code Agent Studio | Synced content, CRM | AI Agent Studio |
| Testing and oversight | Zendesk QA scores AI | Guardrails you set | Simulations, regression | Preview uses sessions | Coverage % and handoff | Governance built in |
| Voice | Voice AI agents | 30 credits per action | Voice listed | Through Freshcaller | Calling in beta | AI voice agents |
| India data storage | No India region | Hyperforce in India | US, EU or Australia | India data centre | No India region | Mumbai and Bengaluru |
| Lock-in and exit | Flows stay in Zendesk | Tied to Data 360 | Desk-agnostic | Freshdesk-bound | HubSpot CRM-bound | Platform commitment |
| Best fit | Resolution-priced AI | Salesforce-run service | Desk-agnostic AI | Freshdesk teams | HubSpot-centred teams | ServiceNow estates |
Honest fit signals — because the fastest way to lose your trust is to pretend one product wins every scenario.
TechBag has no AI customer service guide yet, so Zendesk AI Agents sits outside the category guides. Browse all products to compare it with the rest of the catalogue. →
Drag the sliders (support agents answering customers; support-agent hour cost). Estimates model time spent on repeat questions an AI agent could resolve at an assumed 1.5 hours per agent a year, with 70% of it removed. Both figures are assumptions, and resolution fees are not netted off. Illustrative.
Loaded cost = salary + overheads per productive hour. Illustrative only — your TechBag quote models your actual environment and modules.
Published: AI agents come with every Zendesk Suite and Support plan, carrying 5, 10 or 15 automated resolutions per agent a month by tier (10,000 a year at most); pay-as-you-go resolutions beyond that are US$2.00 each. Plans start at US$19 an agent for Support Team (email only) and US$55 for Suite Team, billed in USD. Forethought on its own is quoted. TechBag forecasts your resolutions first, then quotes in INR with GST.
Best when support already runs on Zendesk
Best for a broader rollout
Best when you keep another help desk
Whatever the list prices above, TechBag negotiates a significantly better deal — with GST-compliant INR invoicing and local support. Ask us for your discounted quote.
Tell us your requirements and current tools — we’ll model it against what you spend today.
Take this into your next vendor call — including ours.
How many routine requests a month could an AI agent fully resolve, and what would the excess cost at US$2.00 each?
Does your tier give 5, 10 or 15 resolutions per agent a month, and how near the 10,000-a-year cap will you run?
Are you on Zendesk, or would Forethought agents on your current desk spare you a migration you do not need?
Which requests need the agent to act — refunds, address changes, order look-ups — and which systems must it reach?
Will you activate the location add-on, and can your DPO accept advanced agents processing data on Google Cloud?
Who reviews AI conversations each week, and will you add Zendesk QA at US$35 an agent a month to score them?
Will you treat Zendesk’s ‘up to 80%’ as a ceiling and set your own target from a pilot on real tickets?
Can finance work with USD, EUR or GBP invoices, or do you need the subscription quoted in INR with GST?
Forecast your monthly automated resolutions first, or let a TechBag advisor scope a pilot on your twenty most common customer requests.
Stats, ratings, review counts and pricing are illustrative and sourced from public materials; verify before purchase.