The significant enhancements to customer self-service will be more impactful and drive significant organizational change, including the reinvention of the role of the agent and new roles to support virtual agents. On average, vendors in this space argue that the number of human agents in the contact center will decrease by 30% in the next two years. Regardless of the specific percentage change in staffing or the specific timeline, big changes are on the horizon.

Organizations can use an agentic conversational AI platform to create simple “tuck-in” self-service applications, but they must think of it as the first step in a multiyear customer service transformation. Agent-powered conversational AI uses modern AI capabilities to follow rules and guidelines to interact with customers and perform tasks on their behalf without requiring step-by-step instruction.

Aisera’s Agentic Platform Intercom. Fin Kore.ai. Kore.ai Agent Platform V1.4.0 LivePerson. Conversational Cloud Netomi. NiCE Cognigy. Cognigy.AI (2025.25 release) Omilia. Omilia Cloud Platform PolyAI. Agent Studio 1.0 Rasa. Rasa Platform Sierra. SoundHound AI. Amelia 7 (7.3.1) Uniphore. Business AI Cloud Yellow.ai. Evaluation Overview We evaluated vendors against three categories: Current offering.

Some assistants act as virtual sales agents, offering personalized product recommendations and guiding customers through applications for savings accounts, loans, credit cards, and investment products.

This means replacing traditional escalation paths with intelligent swarming, proactive automation, and embedded AI agents. Level n support uses AI for IT operations (AIOps), the digital employee experience (DEX), and unified endpoint management (UEM) to detect and resolve issues before they affect users. Level 0 expands self-service through conversational AI and virtual agents embedded in platforms like Teams and Slack.

Its virtual agent enhances user interactions, and its analytics and dashboards provide AI-driven insights. However, its change risk analysis remains limited, relying on questionnaires and offering little visibility into change collisions, and its agentic AI capabilities are still in early stages of development. Customer feedback.

Virtual agent improvement velocity. This measures how quickly the support organization identifies and remediates errors or deficiencies in the performance of virtual agents: It’s the time between when an AI issue is flagged and remediation is deployed. Improvement velocity reflects the agility of your AI lifecycle management and the maturity of your feedback loop processes. Slow virtual agent improvement may indicate bottlenecks in ownership, tooling, or cross-team coordination.

This misplaced confidence, combined with mounting pressure to cut operating costs, will cause companies to deploy genAI-powered chatbots and intelligent virtual agents prematurely — and in contexts where they’re unlikely to succeed. This will erode the customer and brand experience, causing Total Experience Scores to stay flat or decline.

Online chat with a virtual agent (using AI): 6%. Source: Forrester’s Financial Services Banking Survey, 2025.

Include virtual assistants in SLAs. B2B organizations are increasingly using virtual agents to facilitate conversational interactions in real time. When a prospect or customer is ready to be passed to another team member, virtual assistants should follow the same live-team routing processes to ensure smooth transitions and consistent experiences. Any handoffs that include a virtual assistant should have an SLA, regardless of whether the opportunity is moving downstream or upstream.