Encore AI raises $30M to build AI agents that learn from customer calls

Encore AI, a company building artificial-intelligence agents for customer-facing work, has secured $30 million in Series A financing. Team8 led the round for the startup, which is developing software designed to study how businesses communicate with customers and translate those lessons into agents that can handle sales and support activity. The funding gives Encore AI fresh backing as companies test whether automated systems can take on a more central role in conversations that have traditionally depended on human representatives.

The company’s approach begins with records of customer engagement rather than a generic script. Encore AI examines telephone conversations, written exchanges and information held in customer-relationship-management systems. Its stated aim is to locate the methods associated with better results, then incorporate those methods into operational guidance for its AI agents. Those agents may work in tandem with salespeople and support staff, or be used to conduct tasks on their own.

Encore AI’s current direction follows a shift from its original business. Dvir Ginzburg founded the company in 2022 under the name Insait IO. At that stage, it made recommendation technology for financial advisers and relationship managers, a setting where people must draw on client information and determine which next step may be most useful. The company has since adopted the Encore AI name and broadened the underlying idea beyond that earlier financial-services focus.

Under the revised model, the platform reviews exchanges between a company’s staff and its customers. It is intended to distinguish the techniques that helped produce a desired outcome from those that did not. The findings can then be converted into playbooks for the company’s automated agents. This makes the recorded experience of an organization’s own teams, rather than solely a prewritten vendor template, part of the material used to shape how an agent responds.

Ginzburg has described the goal as capturing the best elements of the ways employees already operate. In practice, that means Encore AI is positioning its agents as an extension of established sales and service processes, not simply a replacement for every human interaction. The distinction matters for companies that want automation to reflect their existing customer knowledge, product language and escalation habits while still allowing staff to remain involved when a case requires judgment or follow-up.

The strategy also reflects a broader challenge in customer-service automation: an agent can be fluent without being useful. A system may need access to accurate account context, a clear understanding of the company’s procedures and a reliable way to recognize when it should not proceed alone. By looking at prior interactions and CRM records, Encore AI is seeking to make its software more specific to each customer organization. Whether that produces consistently better outcomes will depend on the quality and relevance of the material being analyzed.

There are limits to what historical interaction data can show. A successful call may reflect an employee’s experience, the timing of an offer, the nature of a particular customer’s problem or factors not captured in a record. The practices identified by a platform therefore still need to be assessed by the businesses using them. Companies will also have to decide how much autonomy to grant an agent in sales and support work, particularly where an inaccurate answer or an inappropriate escalation could affect a customer relationship.

Encore AI’s next stage will put its model to a practical test: turning observed patterns into agents that can operate alongside teams or independently without losing the judgment embedded in human work. The $30 million round provides resources for that effort, but it also raises the importance of demonstrating that the resulting playbooks work across real customer interactions. For businesses considering such tools, the central questions will be how the agents are trained, how their performance is measured and where human employees remain responsible for the final decision.

Source: TechCrunch