The Support Ticket Is Dying and Slack Is Holding the Knife

For about twenty years, "customer support" meant one thing. You had a problem, you opened a ticket, you took a number, you waited. An entire software category - billions of dollars, dozens of public companies - was built on that single metaphor. The ticket was the atom. Everything else was scaffolding around it.

Then the customers left. Not all at once, and not loudly. They just stopped emailing a portal and started typing into the Slack channel they already had open. They pinged a shared Microsoft Teams space. They dropped a message in a Discord server. And they expected the same thing they'd expect from a colleague: a reply in the thread, not a case number in a queue.

The tools that noticed this first are now quietly rewriting the rules of B2B customer support. The ones that didn't are writing press releases about how they're "adding native messaging."

01 / The shiftSupport stopped being a place you go

Here is the number that reframes everything: roughly 80% of B2B interactions now happen digitally, spread across an average of 62-plus touchpoints over six months, according to Forrester's 2025 CX Index. Almost none of those touchpoints begin with someone thinking, "I should file a ticket." They begin in a conversation that's already happening.

That's a problem for software designed around the ticket. When a message lands in Slack and your support tool lives somewhere else, one of two things happens. Either an agent copies the message out - losing the thread, the context, the history - or the message just sits there until it falls through the cracks. Ask any support lead what triggered their last tool evaluation and you'll hear a version of the same sentence: customer messages were getting lost in Slack, with no owner and no way to track them.

So a wave of tools rebuilt support around a different atom. Not the ticket - the account. In this model, the unit of work isn't a stray email; it's a customer company, with all its people, conversations, and history in one place. A Slack thread from that company's engineer isn't pulled out and flattened into a ticket. It stays a thread. It just becomes trackable, assignable, and answerable from one workspace.

"We had to change our workflows to fit the tool, rather than the other way around."A Voltage Park director, on why teams abandon legacy help desks

02 / The upstartTwo designers who deleted the ticket

Plain is the clearest expression of this bet. It was founded in London in 2020 by Simon Rohrbach and Matt Vagni, two designers who had helped scale Deliveroo from roughly 10 people to 2,000. They looked at customer support and saw a ticketing metaphor that hadn't really been questioned since the 2000s. So they questioned it.

Plain organizes everything by account, unifies Slack, Microsoft Teams, Discord, email, and chat into one view, and - crucially for its audience - exposes a public GraphQL API with no rate limits. The same endpoints its own product UI uses internally are open to customers. That's not a footnote; it's the pitch. Plain's buyer is a B2B SaaS company whose customers live in Slack and whose buying committee includes an engineer or a CTO. Teams like n8n, Vercel, Cursor, Sourcegraph, and Raycast run support this way.

In February 2025, Plain raised a $15M Series A led by Battery Ventures, with Index Ventures and Connect Ventures returning, bringing total funding to around $21M. The clearest proof point in its pitch deck is operational, not financial: n8n grew its support volume roughly 20x - from about 100 to 2,000-plus conversations a week - while only doubling its team. The trick wasn't hiring. It was architecture.

03 / The economicsWhy AI is the second half of the story

Rebuilding around accounts and messaging solved the "where." AI is solving the "how much." The math is blunt: an AI-handled support interaction costs roughly $0.50, versus about $6.00 for a human agent. When a routine question can be answered for a twelfth of the price, the pressure to automate the routine isn't ideological - it's arithmetic.

Industry estimates put around 30% of service cases resolved by AI today, climbing toward 50% by 2027. Custom agents on API-first platforms reach 40-60% resolution on the right kinds of queries. The interesting fight isn't whether to use AI - it's whose AI, and how you pay for it.

Three approaches have emerged. There's platform-native AI - Intercom's Fin is the poster child - which sets up fastest but locks you in and charges per resolution (around $0.99 each, which pencils out to roughly $2,000 a month in AI fees alone for a busy 10-person team). There's the AI add-on, like Zendesk's Advanced AI, optional and bolted onto a mature platform. And there's Bring-Your-Own-Agent (BYOA), championed by Plain, where you connect your own models and data. More engineering, more control, no per-resolution tax.

04 / The fieldFour bets on what comes after the ticket

This isn't a one-company story. The whole category is placing bets, and each vendor is really wagering on a different answer to the same question: what is "support" when the ticket goes away?

Front bet on the shared inbox - the cleanest way to run email-heavy support as a team, with collaboration built in. TeamSupport bet on the account, purpose-built for B2B with native Slack and Teams workflows and AI-assisted operations. And the enterprise incumbents - BMC Software and USU - bet on process. They come from the world of IT service management: incident, problem and change management, CMDBs, SLAs, service catalogs. USU has been building this software for over 40 years, older than the internet-era help desk it now competes with. Microsoft, characteristically, bet on distribution - Teams is where a huge share of these conversations already happen.

Every incumbent is now shipping the same feature: "native messaging and AI." The startups didn't add those. They started from them.

05 / The tellNative vs bolted-on

If you want a single test to cut through the marketing, use this one: is the messaging native, or bolted on? When a conversation gets "pulled out" of a Slack thread to become a ticket, it loses its threading and its context - the exact things that made the conversation useful. Critics describe Zendesk's Slack integration as "bolted-on, not native" for precisely this reason. It's a small architectural difference that turns out to decide almost everything downstream: the AI's context, the agent's efficiency, the customer's experience.

The other tell is the API. Newer tools treat deep, no-rate-limit API access as a first-class feature because their buyers increasingly are engineers who want to wire support into Linear, GitHub, and their own internal tools. When developers pick the support stack, the tool with the best API can matter more than the tool with the prettiest inbox.

None of this means the ticket vanishes overnight. Enterprises with genuine process needs - regulated change management, audited SLAs, sprawling asset inventories - still lean on Zendesk, TeamSupport, and ITSM platforms like BMC and USU, and for good reason. But the center of gravity has moved. The default question used to be "where do I file this?" It's becoming "why didn't you just answer me where I asked?" And once customers start asking that, the ticket - the atom that held the whole category together - starts to look less like infrastructure and more like a relic.