Vijay Raina is a seasoned veteran of the SaaS landscape, specializing in the architectural frameworks that power modern enterprise software. His work frequently focuses on how companies can transition from simple workflow automation to sophisticated, data-driven decision-making that mirrors human intelligence. In this conversation, we delve into the evolution of “interaction mining,” a field that is rapidly gaining traction following the successful $30 million Series A funding of Encore AI, led by Team8. We explore the transition from niche financial recommendation engines to a platform that captures the sensory nuances of human speech, the strategic value of mimicking top-performing employees, and the competitive hurdles of taking on established CRM giants who have yet to tap into conversational history as a primary data asset.
The shift from building recommendation software specifically for financial advisers to developing a broader platform centered on “interaction mining” represents a fundamental change in how corporate data is utilized. How does this new approach redefine the value of a company’s historical customer interactions?
This shift essentially reclassifies every single recorded call, email, and text message from a mere compliance archive into a strategic asset that actually trains the next generation of digital workers. By moving beyond simple recommendations, the platform, which was founded in 2022 as Insait IO, now uses that $30 million in fresh capital to build agents that truly understand the flow and rhythm of a successful deal. Instead of just providing static advice to a human, the software mines these interactions to identify the specific moments where a relationship manager’s unique approach led to a positive outcome. This effectively turns a company’s entire conversational history into a repeatable, scalable playbook that can be deployed across the organization, ensuring that the strongest parts of the manual process are never lost. It transforms the way we look at “dead data” by giving it a second life as the primary fuel for autonomous agents.
When we talk about “interaction mining,” the platform is essentially digesting vast amounts of unstructured data from recordings and CRM systems to find what works. Can you explain the specific steps involved in identifying which parts of a complex conversation actually move the needle for a client?
The process is remarkably granular, beginning with the mass collection of varied data streams like call recordings and text messages, which are then meticulously synced with existing CRM information to provide context. The platform divides these customer interactions into distinct stages, allowing the AI to pinpoint exactly where a conversation might be stalling or where a particular employee’s specific tactic excels. By analyzing these stages across more than 40 enterprise customers, the system can detect subtle friction points and inefficiencies that would otherwise remain invisible to even the most seasoned management teams. It isn’t just about identifying the end result of a “won” deal, but about mapping the entire emotional and logical journey to see which specific phrases or structural changes in the conversation lead to a satisfied customer. This allows the AI to learn what works best for any specific client, acknowledging that different employees may be more or less effective at different points in the sales cycle.
One of the more striking aspects of this technology is the claim that AI agents can adopt the specific jokes, anecdotes, and examples used by high-performing human employees. From a design perspective, how does training an agent on human “wit” and personal stories impact the overall effectiveness and authenticity of customer engagement?
Replicating those human quirks is about more than just novelty; it is about building trust and rapport in a way that standard, “robotic” AI has historically failed to achieve. When an agent can weave in a joke or a specific anecdote that a top-performing relationship manager has used effectively in the past, it breaks down the sterile wall usually associated with automated voices. We are essentially packaging the best parts of many different human playbooks into a single, high-performing agent that can communicate via voice or text with a surprising level of empathy. This level of sensory detail helps the interaction feel more authentic to the customer, mimicking the personality of the brand’s most successful individuals. This approach is likely a key factor in why the company has seen its annual recurring revenue grow by more than 5x in the last 18 months, as businesses realize that personality drives conversion just as much as information does.
Despite the rapid growth and new investment from firms like Planven, Lukatz, and Garage, Encore AI is entering a space where legacy giants like Salesforce and HubSpot have a massive footprint. What architectural advantages does a specialized platform have when competing against these established vendors who already hold the keys to so much customer data?
The core advantage lies in the fundamental way the data is structured, as legacy CRM providers typically view customer history as a series of static records rather than a dynamic dataset for agent training. To match this capability, established vendors would likely have to overhaul their entire implementation and technological stacks to prioritize conversational data as the primary building block of their agents. While those giants have the data, they lack the specialized “interaction mining” infrastructure that turns a two-year-old call recording into a functional training manual for an autonomous voice agent. Smaller, more agile firms can focus on making historical conversations the foundation of the agent, rather than just an auxiliary data point. This head start allows them to prove their worth to large financial institutions that are looking for deep, specialized insights and agents that work alongside human teams to recommend real-time responses.
With the success of this recent funding round and the expansion of operations into the U.S. market, where do you see the most significant impact of “interaction mining” occurring within the next few years?
What is your forecast for the future of AI-driven customer interactions?
I forecast a major shift where the distinction between a human-led interaction and an AI-driven one becomes almost entirely blurred, especially in high-stakes environments like finance and insurance. As these agents continue to learn from human playbooks, they will move from being simple assistants to becoming the primary touchpoint for the majority of initial customer inquiries. With revenue growing 5x in such a short window, it is clear that the market is hungry for agents that can actually “think” and “speak” like a company’s best employee rather than a generic chatbot. We will see a future where every interaction is mined in real-time to constantly update a living, breathing digital strategy that eliminates friction and maximizes every sales opportunity. Eventually, the ability to replicate the nuances of human success will be the standard baseline for any company that wants to maintain a competitive edge in customer success and sales.
