Encore AI 完成 3000 万美元 A 轮融资，打造能从客户通话中学习的 AI Agent

Encore AI 是一家研究企业客户互动数据、训练并部署 AI 语音 Agent 的初创公司。这些 Agent 可以与客户支持和销售团队协同工作，也可以自主运行。该公司已完成 3000 万美元 A 轮融资，由 Team8 领投。

该公司由 CEO Dvir Ginzburg 于 2022 年以 Insait IO 之名创立，最初为金融顾问和客户关系经理开发推荐软件。如今，公司已更名为 Encore AI，并将该系统扩展为一个平台：分析企业员工与客户之间的对话，识别哪些方法带来了成功结果，再利用这些发现训练其 AI Agent。

Ginzburg 表示，最终形成的 AI Agent 能够运用企业员工策略手册中最有效的部分。

“有时，我们的 Agent 甚至会讲客户关系经理讲过的笑话，或给出客户关系经理会给出的轶事或案例，因为我们确实是按照那些被证明有效的策略手册运行的……我们构建的 Agent，是由整个流程中许多已经奏效的不同策略手册组成的集合。”他在接受 TechCrunch 独家采访时表示。

Ginzburg 将这一过程称为“交互挖掘”（interaction mining）。公司的平台收集通话录音、电子邮件和短信，并将这些信息与 CRM 系统连接起来。随后，平台会将客户互动划分为不同阶段，并尝试识别对话中哪些部分推动了流程进展、哪些部分未能奏效。

Ginzburg 告诉 TechCrunch，这使 Encore 的 Agent 及其客户能够了解：对于某一特定客户或互动而言，什么方式最有效。因为不同员工在销售或客户成功流程的不同环节，可能分别表现得更有效或更不有效。

该公司的平台还可以帮助企业识别现有客户支持和销售流程中的不足，发现低效环节和摩擦点，并找出关键问题。

Encore AI 交互挖掘工具的截图
图片来源： Encore AI

Encore 表示，其 Agent 可以通过语音或文本直接与客户沟通，也可以充当员工助手，在对话过程中推荐回复内容和沟通策略。

据 Ginzburg 介绍，该公司在全球拥有超过 40 家企业客户，其中大多数是金融机构。他表示，自不到 18 个月前完成种子轮融资以来，Encore 的年度经常性收入已增长超过 5 倍，但拒绝披露确切营收数字或估值。

Encore 较早进入这一市场，但随着 Salesforce、SAP、Zoho 和 HubSpot 等大型 CRM 提供商能够围绕客户数据构建类似的 AI 能力，其市场份额可能会更难守住。不过，Ginzburg 认为，仅仅获得数据访问权限还不够；成熟厂商若要像 Encore 一样，将历史客户对话作为其 Agent 的基础，就必须彻底调整自身流程。

“我们所竞争的最大参与者，并没有将[对话]历史视为他们正在利用的数据点。要让他们开始向现有员工索取对话数据，就需要改变他们整套实施架构和技术架构。”他说。

Planven、Lukatz 和 Garage 也参与了本轮融资，部分银行和保险公司同样参投。Encore 表示，参与本轮融资的一些金融机构在决定投资前，曾先使用过其产品。

这家初创公司计划利用 A 轮融资所得扩大其美国销售业务，并在更多大型金融机构中部署其平台。

Encore AI, a startup that studies companies’ customer interactions to train and deploy AI voice agents that can work alongside customer support and sales teams, or operate autonomously, has raised $30 million in a Series A round led by Team8.

Founded in 2022 as Insait IO by CEO Dvir Ginzburg, the company started out building recommendation software for financial advisers and relationship managers. Now rebranded as Encore AI, the startup has expanded that system into a platform that analyzes conversations between a company’s employees and customers to identify which approaches resulted in successful outcomes, and uses those findings to train its AI agents.

The result, according to Ginzburg, is an AI agent that leverages the strongest parts of the playbooks used by an organization’s employees.

“Sometimes our agents even tell the jokes that the relationship managers are telling, or give the anecdotes or examples that the relationship managers are giving, because we literally run by the playbooks that we see working […] The agent we build is a package of many different playbooks that have worked throughout the process,” he told TechCrunch in an exclusive interview.

Ginzburg calls the process “interaction mining.” The company’s platform collects call recordings, emails, and text messages, and connects that info with CRM systems. It then divides the customer interactions into stages and tries to find out which parts of a conversation helped move the process along, and which failed.

This lets Encore’s agents, and consequently its customers, learn what works best for any particular client or interaction, as different employees may be either more or less effective at different points during a sales or customer success process, Ginzburg told TechCrunch.

The company’s platform also lets companies identify where their existing customer support and sales processes are falling short, identify inefficiencies and friction points, and find key issues.

A screenshot of Encore AI’s interaction mining toolImage Credits: Encore AI

Encore says its agents can communicate directly with customers by voice or text, as well as act as assistants to employees, recommending responses and tactics during conversations.

The company has more than 40 enterprise customers globally, the majority of which are financial institutions, according to Ginzburg. He said Encore’s annual recurring revenue has increased more than 5x since it raised its seed round less than 18 months ago, though he declined to disclose exact revenue numbers or valuation.

Encore’s early to this market, but its share may become harder to defend as large CRM providers like Salesforce, SAP, Zoho, and HubSpot can build similar AI capabilities around their customers’ data. But Ginzburg contends that access to data alone isn’t enough, as established vendors would need to overhaul their processes to make historical customer conversations the foundation of their agents like Encore does.

“The biggest players that we are competing against, they don’t see [conversational] history as a data point that they are utilizing. For them to start asking for conversational data with their current employees will require changing their entire implementation stack and technological stack,” he said.

Planven, Lukatz, and Garage also participated in the round, as did some banks and insurers. Encore said some of the financial institutions participating in the round first used its product before deciding to invest.

The startup plans to use proceeds from the Series A to expand its U.S. sales operations and deploy its platform with more large financial institutions.