From 20% to 70%: How AI Agents Rewrote Support Economics
HubSpot’s Customer Agent shows how quickly AI customer support agents are reshaping support operations. In just twelve months, its autonomous resolution rate jumped from 20% to 70%, with some customers already clearing 85–90%. That means the majority of incoming conversations never reach a human, fundamentally changing how contact centers think about staffing, tiering, and coverage. The product has surpassed 9,000 customers and now consumes over half of all AI credits on HubSpot’s platform, underscoring how central automated support has become. Customers are primarily using it to cover after-hours and weekends and to absorb tier-one tickets so human teams can focus on complex issues. With some organizations rapidly scaling from thousands to hundreds of thousands of monthly AI credits, the data signals a new baseline: an AI-powered contact center where automation is the default, and human agents are deployed selectively for edge cases and high-value interactions.
Beyond Cost Cutting: AI Agents as Revenue and Growth Engines
AI agents are no longer just a way to trim support costs; they are becoming core revenue and growth drivers. HubSpot’s Customer Agent, for example, is central enough that the company has shifted to outcome-based pricing, charging only when an AI agent actually resolves a ticket or delivers a useful sales lead. That move aligns AI value directly with customer service automation ROI and signals confidence in autonomous resolution performance. At the same time, AI features are powering upsell opportunities and deeper platform adoption, as evidenced by rapid growth in AI credit consumption and the introduction of 28-day free trials to accelerate proof of value. This pattern is echoed across the software industry: AI is now embedded in product roadmaps, sales motions, and valuation narratives. Companies that treat AI agents as strategic revenue levers, rather than experimental add-ons, are redefining what a modern support and engagement platform looks like.
Platform Repositioning: Why Software Vendors Are Racing to Brand as AI-First
Vendors such as Amplitude are repositioning themselves as AI-powered platforms to capture the next wave of growth. Amplitude’s leadership describes its mission as building an instrumentation and observability layer that allows companies to track detailed user behavior and feed AI-driven experimentation, guidance, and analytics. Its agreement involving Statsig’s assets and customers expands strengths in data warehouse-based experimentation and feature flagging—capabilities that AI-focused companies already rely on internally. The result is a tighter link between product data, experimentation, and AI agents that can act on those insights. This is not just about feature parity; it is about convincing enterprises that a unified AI-centric platform can replace a patchwork of point solutions. For public software firms, framing themselves as AI-native is also key to justifying long-term growth expectations and responding to investors who increasingly evaluate platforms on their ability to operationalize AI at scale.
The Risk-Reward Equation for Agencies and Enterprise Support Teams
AI agents’ rising autonomous resolution rate creates both opportunity and risk for agencies and enterprises built around traditional support models. HubSpot’s shift to outcome-based pricing and lower AI agent costs triggered a sharp stock reaction, even as revenue and customer counts continued to grow. For the 3,954 agencies in its partner marketplace, the message is clear: value propositions tied purely to manual ticket handling or basic configuration are at risk of becoming the digital equivalent of horse-drawn delivery. At the same time, demand is growing for partners who can design workflows where AI handles most interactions, integrate data sources, and build playbooks for escalation and exception handling. Enterprises face a similar choice. Those who re-architect their support stacks around AI-powered contact center capabilities will free human talent for higher-order work. Those who cling to legacy models may find their economics—and their competitive position—eroding faster than expected.