The support ticket was invented in the 1990s as a digital version of a physical process: a customer writes a letter (the ticket), drops it in a box (the inbox), and waits for someone to pick it up and process it (the agent).

That metaphor made sense when email was the primary support channel. Customers sent one message. Agents processed them in order. First in, first out. Like a mail room.

Thirty years later, we still use the same metaphor. We've added chat, social media, in-app messaging, SMS, and AI, but the underlying model is still "customer creates a ticket, agent resolves a ticket, ticket gets closed."

The ticket-based model is breaking, and AI is what's finally replacing it.

What's Wrong With Tickets

Tickets are one-directional. The customer describes a problem once, at the beginning. Everything after that is a response chain. But problems evolve. The customer discovers new information. The context changes. The rigid ticket structure doesn't accommodate this well.

Tickets create false boundaries. "Is this the same issue as last time, or a new ticket?" Agents ask this. Customers don't care. They have a problem. Whether it's "the same ticket" or a "new ticket" is an internal organizational concept that means nothing to them.

Tickets measure the wrong things. Open tickets, closed tickets, tickets per agent, time to close. These metrics measure throughput, not outcomes. A ticket can be "closed" while the customer's problem persists. It just means someone clicked a button.

Tickets create asymmetric urgency. A ticket that's been open for 5 days looks the same in the queue as one that's been open for 5 minutes. Unless someone manually flags it, age doesn't equal urgency in ticket systems.

What's Replacing It

The shift is from tickets (discrete, bounded events) to conversations (ongoing, contextual relationships).

In a conversation model, when a customer contacts you, their entire history is available. The agent (human or AI) sees every previous interaction, every purchase, every issue. There's no "new ticket." There's just the customer, continuing a conversation that may have started months ago.

Slack adopted this model years ago for internal support. You don't create a "Slack ticket." You send a message in a channel. Someone responds. The conversation is continuous, searchable, and contextual.

Consumer expectations have shifted the same way. When you text a friend about something, you don't create a new thread every time. The conversation is continuous. Customers increasingly expect the same from businesses: message them, get a response, and pick up where they left off.

How AI Accelerates the Shift

AI is what makes the conversation model practical at scale. A human agent can't remember every customer's history. AI can.

When a customer messages "my export is broken again," AI has the context. It knows the customer reported an export issue 3 weeks ago. It knows what the resolution was. It knows the current version of the product. It can respond with: "I see you had an export issue on February 28 that was related to the date format. Is this the same problem or something different?"

That response would take a human agent 5 minutes of reading history. AI does it in milliseconds. The conversation model only works at scale with AI providing the context layer.

Supp's approach is already conversation-oriented. Messages are classified by intent, but they exist within a conversation context. The customer's history, their previous intents, their resolution outcomes, all available alongside the current message. The "ticket" is an artifact. The conversation is what matters.

What Changes Practically

The shift from tickets to conversations changes several things about how support works.

Metrics change. Instead of "tickets closed per day," you measure "conversations with positive outcomes," "issues resolved permanently (not reopened)," and "customer effort across the relationship."

Staffing changes. In a ticket model, agents handle tickets. In a conversation model, agents handle relationships. Some companies are already assigning customers to specific agents (or teams) for continuity. The agent who helped you last time helps you next time. Context carries over. Trust builds.

Tool design changes. Help desks built around ticket queues (Zendesk, Freshdesk) are adding conversation features. Newer tools (Intercom, Front) were built conversation-first. AI tools like Supp are context-first, providing classification and history regardless of the interface.

Customer expectations change. Once customers experience conversation-based support (where the agent knows who they are and what happened last time), they can't go back to "please describe your issue from the beginning."

The Ticket Isn't Dead Yet

To be fair, the ticket metaphor still works for certain support types. Transactional, one-off issues (password reset, order status) don't need a conversation model. They need a fast answer. AI handles these as discrete events, not conversations.

And internally, the ticket is useful for tracking and routing. "Move this ticket to the billing queue" is a practical workflow operation. The internal mechanism can be ticket-based even if the customer-facing experience is conversational.

What's dying is the customer-facing concept of a ticket: the idea that each interaction is isolated, numbered, and closed at the end. Customers don't think in tickets. They think in problems. And problems don't always resolve in a single, bounded interaction.

Where This Is Going

In 5 years, the dominant support model will be proactive and conversational. AI will detect problems before customers report them (monitoring, error detection, usage pattern analysis). When a problem is detected, the system will reach out to the customer, not the other way around.

"We noticed your export failed this morning. We've fixed the issue and re-run the export. Your file is ready: [link]. Sorry about the hiccup."

The customer never submitted a ticket. They never described a problem. They never waited for a response. The "support interaction" was a proactive notification that the problem was already solved.

That's where this is going. The ticket was the customer asking for help. The future is the system providing help before anyone asks.

We're not there yet. But every step toward AI classification, automated resolution, and contextual conversations is a step away from the 1990s ticket model and toward something that actually matches how customers think about support.