Encore AI Raises $30 Million to Train AI Agents From Customer Conversations

Encore AI has raised $30 million in Series A funding to expand a platform that analyses customer interactions and uses the lessons from successful conversations to train AI agents for sales and customer support.

The round was led by Team8, with participation from Planven, Lukatz and Garage, along with several banks and insurance companies. Encore said some financial institutions participating in the investment had initially adopted its technology as customers before becoming investors.

Founded in 2022 by CEO Dvir Ginzburg under the name Insait IO, the company initially developed recommendation software for financial advisers and relationship managers. After rebranding as Encore AI, the startup expanded that technology into a system that studies how employees interact with customers and identifies which approaches produce the best outcomes.

Those findings are then used to create AI agents capable of working alongside employees or communicating with customers independently.

Encore AI learns from successful conversations.

Ginzburg describes thecompany’ss approach as “interaction mining.” Encore collects information from call recordings, emails and text messages and connects those conversations with data stored in customer relationship management systems.

The platform divides interactions into different stages and analyses which parts helped move a customer toward a successful outcome and which approaches were less effective.

Encore then combines successful techniques from multiple employees into playbooks used by its AI agents. That can include specific explanations, anecdotes, examples and even humour that proved effective during previous customer conversations.“

"The agent we build is a package of many different playbooks that have worked throughout the process," Ginzburg said.

The approach is designed to avoid simply copying one high-performing employee. Different workers may perform better during different stages of a sales or customer support process, allowing Encore to combine the strongest elements from across an organisation.

AI agents can assist workers or operate independently

Encore’s agents can interact directly with customers through voice or text. They can also operate as assistants to human employees, suggesting responses or recommending tactics while conversations are taking place.

The underlying analysis can provide companies with information beyond agent training. Encore says businesses can use the platform to identify inefficiencies in their existing sales and support operations, uncover friction points and determine where customer interactions are breaking down.

The startup currently has more than 40 enterprise customers worldwide, with financial institutions representing the majority of its client base.

Ginzburg said Encore’s annual recurring revenue has grown by more than five times since its seed funding round less than 18 months ago. The company has not disclosed its current revenue or valuation.

Large CRM providers could become competitors.

Encore is entering the AI agent market as major enterprise software providers increasingly add similar capabilities to their platforms. Companies including Salesforce, SAP, Zoho and HubSpot already manage large amounts of customer data that could potentially be used to build competing AI products.

Ginzburg argues, however, that simply possessing customer data does not replicateEncore’ss approach. He says established CRM companies generally do not treat the history of employee-customer conversations as the primary foundation for training their agents.

Building systems around that conversational history, he argues, could require significant changes to how those platforms collect information and structure their technology.

Encore plans to use its new funding to expand sales operations in the United States and deploy its platform across more large financial institutions.

As companies increasingly experiment with AI agents for customer-facing work, Encore is betting that agents trained on what an organisation’s best employees have already proven works will be more effective than systems built from generic instructions alone.