Why Customer Support Software Is Having an Identity Crisis

For twenty years, customer support software agreed on what it was. You had an inbox. Messages became tickets. Tickets got a number, a status, and a human. The whole category - from the earliest web helpdesks to the biggest enterprise suites - was really one idea dressed six ways: route the problem to a person, then measure how fast that person typed. In 2026, that idea is quietly falling apart, and the companies selling support software cannot agree on what should replace it.

The tell is in the pitch. Ask any vendor what they sell now and you will hear the same sentence, almost word for word: support that resolves itself. SAP says it. ServiceNow says it. Gorgias, Kayako, Supportbench and Thena all say it. The sentence is doing a lot of work, because underneath it sit six very different companies making very different bets. Some are genuinely rebuilding what a support tool is. Some have taken the same old inbox and painted a chatbot on the front. The interesting question is not who has AI. Everyone has AI. The question is who lets that AI actually do something.

The giants are automating from the top down

Start with the largest players, because they have the most to lose. ServiceNow did not begin life as a support company. It grew up in IT service management - the ticketing system your own colleagues use when their laptop dies - and spent years expanding sideways into customer service, HR and everything else a big company runs on workflows. In 2025 it went shopping, buying Moveworks, Logik.ai and CueIn, and in 2026 it launched what it calls an "Autonomous Workforce": thousands of preconfigured AI agents that sit inside those workflows and close cases without waiting for a human to pick them up.

SAP is playing the same game from a different corner of the enterprise. Its AI layer, branded Joule, threads through Service Cloud and the rest of its sprawling estate. For a company already standardized on SAP or ServiceNow, this is the path of least resistance - the AI shows up where the data already lives, and support becomes one more function the platform automates. The catch is the flip side of that strength. When support is one workflow among a hundred, it rarely feels built for support. It feels built for the org chart.

The specialists are rebuilding from the customer up

The more interesting movement is happening below the giants, where smaller companies picked one type of customer and refused to serve anyone else. Gorgias is the clearest example. Founded in 2015, it bet entirely on ecommerce, wiring itself so deeply into Shopify that it became an official Premier Partner for customer experience. More than 17,000 brands run it. What makes Gorgias worth studying is not the integration count - it is the philosophy. Its AI Agent does not stop at answering "where is my order?" It edits the subscription, issues the refund, updates the shipping address. Support here is not a conversation. It is a task that gets finished.

Thena made a smaller, sharper bet: that B2B customers do not want to open a ticket portal at all. They already live in Slack, so support should meet them there. Thena was built Slack-first - with Microsoft Teams, in-app chat and email alongside - and its AI does not just draft replies. It runs the actual B2B jobs behind them: provisioning, renewals, reporting. The company's own line is that a single support lead can now manage hundreds of customers across channels, work that used to demand a four-person team. Vercel, ClickHouse, Amplitude and FOX are on the customer list, which is a decent proxy for whether the Slack bet is real.

Then there is Supportbench, the least famous name here and, in a way, the most instructive. Founded in Vancouver in 2015, it has raised only about $1.5M total - a number the enterprise players would lose in the sofa. It spent a decade on the unglamorous plumbing of B2B support: account history, dynamic SLAs, escalation workflows, customer health scoring, predictable per-agent pricing. For years that work looked boring. Then the AI wave arrived and made every bit of it suddenly matter, because an AI agent is only as good as the account context you can hand it. Supportbench's whole pitch is that depth beats size.

What connects these three is a refusal to be everything to everyone. Gorgias would be a poor fit for a bank; Thena would be strange for a consumer app with a million anonymous users; Supportbench is not trying to win the ecommerce crowd. That narrowness is the point. When you know exactly who your customer is, you can wire the AI to take the specific actions that customer needs - a refund, a renewal, a provisioning request - instead of offering a generic assistant that can only summarize. The specialists are not smaller versions of the giants. They are a different shape of product, built outward from one type of customer rather than downward from an org chart.

Where the six players sit

The legacy question: what happens to Kayako?

Kayako has been doing helpdesk since 2001, back when "web-based support software" was itself the disruptive pitch. Its shared inbox and live chat defined what a lot of teams still picture when they hear the word helpdesk. Watching a tool like this adapt tells you more about the market than any startup deck, because legacy products cannot rip out their foundations. They can add a chatbot. They can add suggested replies. What they struggle to do is change the underlying assumption - that a ticket is a thing a human answers - without becoming a different product. That is the quiet trap the whole category is walking toward.

Follow the pricing, not the demo

If you are buying, ignore the demo for a minute and look at the price sheet. It leaks the honest self-image. Supportbench charges around $32 per agent per month - classic tool pricing, you are renting seats for humans. Gorgias runs roughly $60 to $900 a month by tier, then adds about $0.90 to $1.00 per AI resolution. That per-resolution line is the whole story in one number: it only makes sense if the software is doing work a human would otherwise do. When a vendor charges for outcomes instead of seats, they are betting their own margin that the AI resolves. That is a very different promise from "unlimited agents, now with a chatbot."

Product	Best for	Starting price	Model
Gorgias	Ecommerce / Shopify	~$60/mo + ~$0.90-$1 per resolution	resolve
Thena	B2B SaaS in Slack	Quoted	resolve
Supportbench	Complex B2B support	~$32/agent/mo	route+
ServiceNow	Large enterprise	Enterprise contract	resolve
SAP	SAP-standardized orgs	Enterprise contract	route+
Kayako	General shared inbox	Quoted	route

How "resolve-first" is each one, roughly

Gorgias

action-taking

Thena

runs B2B tasks

ServiceNow

autonomous agents

SAP

Joule-assisted

Supportbench

deflect + context

Kayako

assisted inbox

What the crisis is really about

Put the six side by side and the split is obvious. On one side are tools that still measure success in tickets a human closed quickly. On the other are tools that measure success in tickets no human ever touched. The category is not fragmenting because the vendors disagree about AI. They fragment because they disagree about what support is for. Is it a queue to be worked, or a set of outcomes to be delivered? The old software was built for the first answer. The new software is being built for the second, and it does not fit inside the old shape.

This is why the market feels unsettled rather than settled. In most software categories, a wave of AI would produce a clear winner and a lot of copycats. Here it has done the opposite. It has pulled the vendors apart, because the moment support stops meaning "a queue humans work" and starts meaning "outcomes software delivers," the right design depends entirely on which outcomes and whose customers. A tool tuned to resolve Shopify refunds and a tool tuned to run enterprise IT cases share a slogan and almost nothing else. The identity crisis is not weakness. It is what a category looks like mid-rebuild, before anyone agrees on the new definition.

For anyone buying, the practical advice is short. Match the tool to your customer, not to the logo. Selling to Shopify shoppers, use Gorgias. Supporting B2B customers who live in Slack, look hard at Thena. Running deep, high-touch enterprise accounts, Supportbench and the big platforms earn a look. And whatever the salesperson says, ask the one question that cuts through the whole identity crisis: when your AI meets a real customer problem, does it solve it, or does it just write you a nicer note about it?