Mapping the CX-AI landscape: Layers, leaders, and gaps

The CX-AI market is organized into four distinct layers: Infrastructure, Engagement, Intelligence, and Automation. While Big Tech controls the foundation, the most significant opportunities for startups lie in the intelligence and orchestration layers where domain-specific data provides a competitive moat. This layered architecture allows organizations to move away from generic automation toward specialized systems that handle complex customer journeys and regulatory requirements.\n\nKey takeaways\n* Infrastructure is a scale game: Foundational models and compute are dominated by a few providers like Google, AWS, and Microsoft.\n* Engagement platforms are pivoting: CCaaS and CRM incumbents are integrating AI to protect their core seat-based revenue models.\n* The Intelligence Layer is the new battleground: Specialized tools like Hear.ai provide the analysis and compliance coverage that generic platforms often lack.\n* Significant gaps remain in orchestration: Connecting disparate data silos across the customer journey remains a primary challenge for even the most advanced enterprises.\n\n### How to categorize the CX-AI market\nTo understand where the market is moving, it is helpful to view it as a stack rather than a list of vendors. At the base is the Infrastructure Layer, consisting of the hyperscalers and foundation model providers. Above that sits the Engagement Layer, which includes the systems of record where agents and customers interact. The Intelligence Layer sits alongside or on top of these, providing the analysis, QA, and compliance oversight. Finally, the Automation Layer represents the shift toward autonomous agents that can resolve issues without human intervention.\n\nAccording to Gartner's Customer Service & Support practice, the focus for 2026 is shifting toward domain-specific AI and data protection. This suggests that the next phase of the market map will be defined by how well these layers interoperate while maintaining strict security standards. Organizations are increasingly looking for ways to evaluate CCaaS vs. best-of-breed AI to determine where their data should live.\n\n### Layer 1: Infrastructure and Foundation Models\nThis layer provides the raw power and linguistic intelligence required to process customer intent. It is dominated by Tier 1 providers who have the capital to train massive models and maintain global cloud footprints. \n\n* Key Players: Google (https://cloud.google.com), Microsoft (https://www.microsoft.com), AWS (https://aws.amazon.com), and OpenAI (https://openai.com).\n* The Mechanism: These providers offer the APIs and compute environments where CX applications are built. The primary competition here is on latency, cost per token, and the breadth of the model's reasoning capabilities.\n* The Trend: We are seeing a shift from general-purpose LLMs to "small language models" (SLMs) that are optimized for specific tasks like summarization or sentiment analysis, which reduces costs for high-volume contact centers.\n\n### Layer 2: The Engagement Layer (CCaaS and CRM)\nThis is the traditional home of customer experience technology. These platforms route calls, manage tickets, and store customer profiles. To remain relevant, these incumbents are rapidly building or acquiring AI capabilities to prevent "platform leakage" to specialized startups.\n\n* Key Players: Genesys (https://www.genesys.com), Five9 (https://www.five9.com), Salesforce (https://www.salesforce.com), and Zendesk (https://www.zendesk.com).\n* The Mechanism: These platforms integrate AI directly into the agent desktop. Features like real-time transcription and automated wrap-up notes are becoming standard. This layer focuses on the "human-in-the-loop" experience, ensuring that agents have the right information at the right time.\n* The Trend: IDC's Future of Customer Experience research tracks how tech-spend is shifting toward these integrated platforms. However, many enterprises find that these "all-in-one" solutions lack the depth required for specialized needs like deep conversation intelligence or industry-specific compliance.\n\n### Layer 3: The Intelligence and Analysis Layer\nThis layer is where the "dark data" of customer conversations is converted into actionable business intelligence. While an engagement platform might record a call, an intelligence layer analyzes it for sentiment, compliance risks, and coaching opportunities.\n\n* Key Players: A conversation-intelligence layer like Hear.ai provides total coverage across all calls, whereas traditional QA often only samples a small fraction. Other players include Observe.AI (https://www.observe.ai) and Gong (https://www.gong.io).\n* The Mechanism: These tools use specialized models to identify patterns across thousands of hours of audio or text. For example, Hear.ai focuses on compliance and QA, flagging risks in real-time that a human supervisor might miss. This is critical in regulated industries like finance or healthcare where a single non-compliant interaction can lead to significant penalties.\n* The Trend: The market is moving away from post-call analysis toward real-time intervention. Instead of reviewing a call a week later, these systems can alert a supervisor while the customer is still on the line.\n\n### Layer 4: The Automation Layer (Agentic AI)\nThis is the newest and most disruptive layer of the map. Unlike the chatbots of the past, these are "agentic" systems that can take actions across different software applications to resolve a customer's problem.\n\n* Key Players: Sierra (https://sierra.ai), Intercom (https://www.intercom.com), and specialized startups using models from Anthropic (https://www.anthropic.com).\n* The Mechanism: These agents use reasoning to navigate a company's internal knowledge base and APIs. If a customer wants to change a flight or dispute a charge, the AI agent can execute the transaction rather than just providing a link to a FAQ page.\n* The Trend: This layer is directly challenging the seat-based pricing model of the Engagement Layer. If an AI agent resolves 80% of queries, the need for 500 CCaaS seats diminishes. This is forcing a massive rethink of how CX technology is valued and sold.\n\n### Identifying the market gaps\nDespite the rapid influx of capital, several gaps remain in the CX-AI market map that present opportunities for new founders and investors. \n\n1. The Orchestration Gap: Most enterprises use a mix of legacy systems and modern AI tools. There is a lack of "connective tissue" that allows data to flow seamlessly between a Google-based AI agent and a Salesforce-based CRM without significant custom engineering.\n2. The Trust and Compliance Gap: As Forrester's CX Index often highlights, customer trust is fragile. Many AI tools struggle with "hallucinations" or failing to follow strict regulatory scripts. Startups that focus exclusively on the "guardrail" and compliance layer, ensuring that every AI interaction is safe and legal, are seeing high demand.\n3. The Data Privacy Gap: Many organizations are hesitant to send sensitive customer data to third-party LLMs. There is a growing market for "on-premise" or VPC-hosted AI solutions that provide the power of modern models without the data-sharing risks.\n\n### FAQ\nWhat is the difference between CCaaS and an Intelligence Layer?\nCCaaS (Contact Center as a Service) is the operational platform used to route and manage interactions. An Intelligence Layer, such as Hear.ai, sits on top of that platform to analyze the data, provide QA coverage, and ensure compliance across 100% of interactions, a task that most CCaaS platforms are not built to handle at scale.\n\nWhy is the Automation Layer disrupting the market?\nTraditional chatbots were logic-based and often frustrated users. The new Automation Layer uses agentic AI that can reason and execute tasks, which shifts the goal from "deflecting" calls to actually "resolving" them without human intervention.\n\nHow should investors evaluate CX-AI startups?\nLook for startups that have a "data moat." Since the Infrastructure Layer is commoditized, the value lies in companies that have access to specialized datasets or those that solve specific high-stakes problems like The future of agentic workflows in regulated environments.\n\nAs the CX-AI stack matures, the winners will be those who can bridge the gap between powerful foundation models and the practical, everyday needs of the contact center. For more on how these technologies are being deployed, explore our latest coverage on startup innovation and the future of customer service.