Typewise launches Nova to run customer-service AI agents

Typewise positions Nova not as another customer-facing chatbot but as an orchestrator sitting above a company's customer-service and business applications. When a request arrives, an AI supervisor identifies intent and routes the case to a specialist agent that pulls information from some of the more than 3,500 connected knowledge bases and systems, such as CRM, ERP, billing, commerce and logistics platforms. Companies authorize connections and describe in plain language what the AI should do, avoiding the need to program every workflow, and administrators can set automation limits and require approval for sensitive actions.

The company's own research indicates that most customer service teams run AI as disconnected tools, and Typewise argues that the operational work required to keep AI systems accurate and current is preventing many pilots from reaching production. Chief Executive and co-founder David Eberle framed this as a new operations job created by every AI agent deployment, with someone needing to monitor agents, update knowledge, test changes and decide when they are ready for more responsibility. Nova is presented as that 24/7 operations team. Early customer results cited by Typewise include HealGreen LLC, where agents handle 70% of incoming inquiries and fully resolve between 75% and 85% of them, and Beurer North America LP, which resolves 65% of inquiries with a success rate above 90% and cut average human handling time from 12 minutes to six.

The stakes hinge on whether that operations burden is real enough to pay for, and on how quickly customers move beyond an initial use case that Typewise says can go live within minutes to deployments that typically expand over several days. Nova drafts replies alongside human employees first, and the outcome-based pricing means unresolved requests generate no revenue, so the test is whether partial resolutions and the promotion ending today convert into longer-term expansions.