Customer Support Software Under the AI Wave: What the Data Shows

Why customer support software is the front line of the AI wave

Customer support was the first enterprise workflow that large language models could plausibly automate end to end. That made support software both the fastest-growing AI category and the most crowded one. WillItSurvive's assessment of 10,525 companies assessed in July 2026 includes a dense cluster of support-adjacent vendors inside its AI vertical, and the pattern there is uncomfortable for anyone building or joining a standalone support-bot company.

Of the 263 AI companies and labs assessed in July 2026, 170 were rated losing its protection and only 2 were judged hard to displace. The support software names sit squarely inside that squeezed middle, exposed to the same forces dragging down the broader AI vertical.

Where the support vendors stand

The assessment tracks a long list of conversational AI and customer-experience companies. Here is how the July 2026 assessment placed the most relevant ones:

LivePerson: assessed as little protection left, the most exposed of the named support vendors in the vertical.
Sendbird: Losing its protection. The assessment notes it is losing AI customer-service deals to Salesforce Agentforce, Zendesk, and Microsoft Copilot Studio even as it pivots from chat APIs.
Yellow.ai: Little protection left. Boost.ai: Losing its protection.
Ada, Forethought, Kore.ai, Rasa, Observe.AI: all losing its protection.
Voiceflow and Aisera: Protection under test.
Xiaoi Robot, a Chinese conversational AI veteran: Little protection left, squeezed by Chinese hyperscalers Alibaba, Baidu, Tencent, and Ant Group, which bundle cheaper foundation models such as Qwen, Ernie, and DeepSeek directly into enterprise platforms.
Cresta and PolyAI (agent-assist and voice): Losing its protection.
Decagon AI, a newer AI support agent company: Losing its protection, though among the steadier names in this group.
Moveworks (employee support rather than customer support): Defensible for now, one of the stronger positions in the cluster.
Cognigy: Defensible for now.

Among the customer support and conversational AI vendors named in the July 2026 assessment, LivePerson stood as wide open, the most exposed of the group, while Moveworks and Decagon AI held comparatively steadier ground.

The mechanism: platforms shipping support AI natively

The sector take from the July 2026 AI vertical assessment is blunt: value that lives as a thin layer on top of general models tends to get absorbed, while value below or around the model layer holds up better. Support software is almost entirely above the model layer. A support bot's core capability, understanding a customer message and drafting a resolution, is now a native feature of the platforms where support already happens.

The assessment pressure map documents this directly. Sendbird's losses trace to Salesforce Agentforce, Zendesk, and Microsoft Copilot Studio, three platforms that own the ticket, the CRM record, or the enterprise contract already. Chinese support and voice vendors like Xiaoi Robot, AISpeech, and CloudWalk face the same dynamic from Alibaba, Baidu, Tencent, and Ant Group, which bundle cheaper foundation models such as Qwen, Ernie, and DeepSeek directly into enterprise platforms.

What separates the stronger positions

The support vendors under less pressure share a pattern. Moveworks, defensible for now, sells into internal employee support, a workflow tied deeply into IT systems rather than a public chat widget. Decagon AI and Cresta, both losing its protection, sell workflow depth and agent-assist rather than a bare chatbot. Cognigy, defensible for now, is one of only a few in the cluster rated above shaky ground.

Compare that with LivePerson, rated little protection left, or Yellow.ai, also wide open. The assessment's broader finding for the AI era is that companies whose product can be replicated by a model upgrade or a platform feature carry the most exposure, and pure conversational front-ends fit that description more tightly than any other support sub-segment.

The assessment also records what happens at the end of that road elsewhere in the AI vertical: Humanloop's team was absorbed by Anthropic, Rewind AI was acquired and shut down by Meta, Covariant became a shell after Amazon hired its founders in 2024, and Phind shut down in January 2026 after ChatGPT and Gemini absorbed its niche. None of these are support companies, but they are documented in the same July 2026 assessment as the pattern support vendors are trying to avoid: acquihire or absorption rather than independent scale.

Retail is where the demand side shows up

Support software's customers are mostly large consumer-facing companies, and the retail vertical in the same assessment shows why buyers are consolidating rather than adding tools. Of the 206 retail and consumer conglomerates assessed in July 2026, 95 were rated hard to displace, a far higher share than in the AI vertical, where pressure runs noticeably higher across the board. Big retailers like Walmart, Woolworths, and Reliance Retail have the scale to standardize on one platform vendor's bundled AI support rather than stitch together point tools. That buying behavior is exactly what pushes standalone support vendors toward the shaky end of the distribution.

Questions this data answers

Is standalone customer support software losing ground to platform vendors? The July 2026 assessment records specific losses: Sendbird was noted as losing AI customer-service deals to Salesforce Agentforce, Zendesk, and Microsoft Copilot Studio. Most named support vendors (Ada, Forethought, Kore.ai, Rasa, Yellow.ai, Boost.ai) were rated losing its protection or worse.

Which customer support AI companies looked strongest in July 2026? Moveworks and Cognigy were the only support-adjacent names rated defensible for now. Decagon AI, though losing its protection, was among the steadier names among newer support-agent startups.

Which support software company faced the most pressure? LivePerson, rated little protection left in the July 2026 assessment, stood as the most exposed among the named support vendors. Xiaoi Robot followed close behind, squeezed by Chinese hyperscalers bundling cheaper foundation models into enterprise platforms.

Does this mean these companies are finished? No. Standings describe evidence as of July 2026, not a prediction of what happens to any individual company next. WillItSurvive's standings separate stressed companies from stable peers based on the facts on record at the time, not a forecast of what any one firm will do.

What should someone watch to track this category? Two dated signals from the assessment: whether more AI tooling and wrapper startups get quietly acquihired by OpenAI, Anthropic, Google, or Meta rather than raising new rounds, and whether Chinese hyperscalers keep launching bundled enterprise AI platforms. Both would confirm the thin layer above foundation models continuing to collapse into the platforms beneath it.

Watch-items, dated

Check whether Sendbird announces further pivots or customer losses to Salesforce, Zendesk, or Microsoft through 2026-2027; the July 2026 assessment flagged this as the live pressure.
Check LivePerson's next reported results against its little protection left standing from July 2026.
Watch whether Moveworks and Cognigy hold their better-than-category standings as platform vendors extend into internal support.

Assessments reflect evidence available July 2026. Companies change; treat watch-items as things to verify, not conclusions.