AI Agent for Customer Service and Support
Quickchat AI resolved 91% of customer questions without a human across all customers in 2026. Enterprise pricing starts at $0.50 per resolved conversation, and the agent works inside the helpdesk you already use.
Free plan · No credit card required
91% resolved by AI
Where's my order? It's been 6 days.
Found order #5120 in Shopify
It's out for delivery and should arrive today. Here are the details:
The rest go to your team, with the context
I was charged twice for my subscription.
I can see two charges on your account. Our billing team needs to check them, so let me connect you.
Anna from Billing joined
A Anna
Hi, I can see both charges. I've refunded the duplicate, and it will show on your card in 3 to 5 days.
“Quickchat AI has dramatically improved our customer engagement by ensuring we are always available, regardless of time zones.”
Trusted by
Used by
How the 91% resolution rate is measured
Quickchat AI's resolution rate across every customer in 2026, next to the average Intercom publishes for Fin.
Resolution rate in 2026 to date
Each point: all customer conversations from March up to that month
| Month | Quickchat AI |
|---|---|
| March 2026 | 90% |
| April 2026 | 90% |
| May 2026 | 91% |
| June 2026 | 91% |
| July 2026 | 91% |
| August 2026 | 91% |
| September 2026 | 91% |
91%
of customer conversations resolved by AI in 2026 so far. Intercom's published average for Fin is 76%.
* Resolution rate: conversations the AI resolved without a human, out of all conversations that contained a question, March to September 2026. Small talk is not counted.
- Confirmed resolution
- After the AI's last answer, the customer confirms it helped ("Ok thanks", "That helped") or through the rating survey.
- Assumed resolution
- The customer ends the conversation without asking for a person.
- Not a resolution
- The conversation is handed off to your team, or the AI only answered a greeting.
- Not counted
- Small talk with no question. It appears in Insights but is left out of the rate.
What a resolved conversation costs
Billed per resolved conversation on Enterprise. Conversations handed off to your team cost nothing.
Enterprise plans are billed per resolution, from $0.50, with a minimum of 2,000 resolutions a month and lower rates at higher volume. A conversation is billed only when it meets the resolution definition above, so handed-off and unresolved conversations cost nothing.
Example: 10,000 customer questions a month
| Per month | Quickchat AI | Intercom Fin |
|---|---|---|
| Resolution rate (published average) | 91% | 76% |
| Resolved by AI | 9,100 | 7,600 |
| Price per resolution | $0.50 | $0.99 |
| AI cost | $4,550 | $7,524 |
| Left for your team | 900 | 2,400 |
Illustrative. Uses each vendor's published average resolution rate and list price as of June 2026. Your rate depends on your knowledge base and the questions your customers ask.
$0.50
Self-serve plans are flat monthly subscriptions from $9 to $999, with a free plan for testing. For comparison, Intercom lists Fin at $0.99 per resolution, and Salesforce lists Agentforce at $2.00 per conversation whether or not it is resolved.
- Quickchat AI $0.50
- Intercom Fin $0.99
- Agentforce $2.00
- Enterprise: from $0.50 per resolution, 2,000 resolutions a month minimum, volume discounts.
- Self-serve: free plan, then monthly plans from $9 to $999.
- Not billed: conversations handed off to your team, unresolved conversations and small talk.
What the agent does in a support conversation.
It answers, takes actions, hands off and shows its work.
Human handoff
Hands off when it should. On rules you write, with a summary for your teammate.
The AI Agent passes a conversation to a person when it detects that it cannot resolve the issue, when the customer asks for a human, or when the conversation matches a handoff rule. Rules are written in plain language per topic (for example "refund requests over $200" or "legal complaints"), and a keyword list triggers handoff immediately when a customer uses one of those words.
Before transferring, the agent can ask the customer to confirm, using wording you provide. Outside the working hours you set, it tells the customer when to expect a reply. Your teammate receives the conversation with an AI summary, and the inbox records the reason for each handoff. In Intercom and Zendesk, the conversation is passed to a human inside that helpdesk.
- Topic-based routing sends each handoff to the team member whose description matches the conversation topic.
- Email notifications tell your team when a conversation needs a person.
- Insights tracks the handoff rate, so you can see which topics the agent still escalates.
Website chat
I'd like a refund for my $340 order, it arrived damaged.
I'm sorry it arrived damaged. Refunds over $200 are handled by our team. Shall I bring someone in?
Yes, please.
Sarah from Support joined
AI summary for Sarah
- Order #4821, $340, arrived damaged
- Customer asked for a refund
- Rule: refund requests over $200
Actions
Does the task, not only the answer. Through Shopify, your API or an MCP server.
Actions let the agent look things up and make changes during a conversation. A Shopify order lookup needs the order number and the customer's email. HTTP request actions call your own API with the details the agent collects, and MCP servers give it tools from other systems.
The agent only runs the actions you turn on, and each one appears in the conversation's trace.
- Order status from Shopify
- HTTP requests to your own API
- Tools from an MCP server
Website chat
Can I move my delivery to Saturday?
Found order #4821 in Shopify
Checked free slots in your delivery API
Booked Saturday, 10:00 to 12:00
Done. Your delivery is now on Saturday between 10:00 and 12:00.
Answer trace
Shows why it said what it said. For every reply.
Open any conversation to see the reasoning behind each reply, the knowledge sources the agent read and the actions it ran.
When an answer is wrong, the trace shows whether the source was missing, outdated or misread, so you fix the source rather than guess. Run the same question again after the fix to confirm the new answer.
- Reasoning for each reply
- Knowledge sources it read
- Actions it ran
Website chat
Can I return a sale item?
Yes. Sale items can be returned within 14 days for store credit.
Why AI said that
- Source
- Returns policy, section 3
- Reasoning
- Sale items: 14 days, store credit
- Action
- None needed
Helpdesk and channels
Works inside the helpdesk you already use. No migration.
In Intercom, the agent replies in your inbox as a teammate and answers assigned conversations or all of them. In Zendesk, it can reply to every new conversation or only the ones you route to it. In HubSpot, it answers live chat, email and Messenger in HubSpot Conversations and creates contacts and tickets as it works.
Handoff happens inside the same tool, so your team keeps working where it already does.
- Intercom, Zendesk and HubSpot
- Website chat, email, WhatsApp, Messenger, Instagram, Telegram and Discord
- One agent and one knowledge base across every channel
Security and privacy
Customer data stays yours. Scrubbed before the model sees it.
Personal details such as card numbers and email addresses are removed before a message reaches the model, and conversations are not used to train AI models. GDPR is in place, EU data hosting is available, and SOC 2 and ISO 27001 are in progress.
Team roles (Administrator, Editor, Support and Chat User) decide who can change the agent and who only answers conversations.
- Personal details removed before the model
- Conversations not used for training
- EU data hosting available
Maybe* Tech: 93% of inquiries handled by AI, inside Intercom
Maybe* Tech connected Quickchat AI to its existing Intercom setup instead of replacing it. The AI Agent now handles more than 600 inquiries a day, and the average first response time went from hours to 13 seconds.
- inquiries a day
- 600+
- handled entirely by AI
- 93%
- average response time, down from hours
- 13 s
“One of the key reasons we chose Quickchat AI, other than their great team, was that we wanted an AI that we could integrate with our existing business systems, particularly Intercom.”
Polly Barnfield
CEO, Maybe* Tech
Frequently Asked Questions
Got more questions? Contact us