Encore AI raises $30 million to turn top sales calls into agent playbooks
Dvir Ginzburg is taking a recommender-system thesis into enterprise voice agents, with Team8 leading the Series A and customers joining as investors.
By Ryan Merket · Published
Why it matters
Encore AI is betting that enterprises' strongest AI advantage is their own conversation history. Customer-investors pair financing with experience using the product inside financial institutions.
Encore AI founder Dvir Ginzburg has raised a $30 million Series A to build AI agents trained on the calls, messages and sales tactics that already work inside large enterprises. TechCrunch reported on July 29th that Team8 led the financing, with Planven, Lukatz and Garage participating alongside banks and insurers.
Ginzburg brings an unusually research-heavy background to a category dominated by sales automation pitches. He studied computer science and electrical engineering at Tel Aviv University, continued into geometric machine learning and doctoral work, and worked on recommender systems at Microsoft after engineering roles at Meta and Cisco. Encore AI is the commercial extension of that path: take a large set of past choices and outcomes, find the patterns that matter, and recommend the next action.
Ginzburg founded the operation in 2022 as Insait IO. Its first product recommended financial products and actions to advisers and relationship managers. The later rebrand to Encore AI accompanied a larger bet: the same underlying approach could study an enterprise's customer conversations, determine how its strongest employees move people through a funnel, and encode those tactics into software that can coach workers or speak directly with customers.
Mining the conversations enterprises already own
Encore AI calls its approach "interaction mining." The platform collects call recordings, chats, emails, funnel data and CRM records, then divides customer journeys into stages and looks for tactics associated with successful outcomes. According to Ginzburg, those patterns can include the timing of a follow-up, the handling of an objection, an anecdote or even a joke that repeatedly works for a human representative.
"The agent we build is a package of many different playbooks that have worked throughout the process," Ginzburg told TechCrunch.
Ginzburg's pitch centers on learning from how high-performing employees behave, including the informal techniques that never make it into a manual. Encore AI aims to turn those observed tactics into playbooks for its agents.
Encore AI says its software can operate in two modes. Wingman and Autopilot respectively coach employees during live conversations and run customer journeys autonomously over voice or text. Encore AI also describes its interaction-mining technology as patented. Those product and patent claims come from Encore AI's materials.
The approach creates a demanding data problem. A tactic that correlates with a sale may not have caused it, and the best move can vary by customer, product, channel and regulatory constraint. Historical conversations can also encode poor habits alongside useful ones. Encore AI's value will depend on whether it can separate repeatable techniques from coincidence while giving customers enough control to prevent an agent from copying behavior that should have stayed human.
Customers are also underwriting the bet
Encore AI has more than 40 enterprise customers globally, most of them financial institutions, Ginzburg told TechCrunch. He also said annual recurring revenue had grown more than fivefold since Encore AI's seed round less than 18 months earlier. Encore AI did not provide an exact revenue base, so the growth figure shows momentum without establishing the scale of the business.
Some banks and insurers in the Series A had used Encore AI before investing, according to TechCrunch. That customer-investor overlap gives Ginzburg more than a financing endorsement. Financial institutions can supply the complex, regulated deployments Encore AI needs to refine its product, while their investment gives them a stake in software that could reduce staffing pressure and improve conversion across large volumes of customer interactions.
It also concentrates Encore AI's early progress in a sector where adoption moves slowly and mistakes carry consequences. Financial institutions need agents that can follow disclosure requirements, protect sensitive data, escalate unusual cases and leave an auditable record. A system that can learn from successful human conversations while respecting those boundaries would have a credible route into other regulated industries. Failures in those controls would limit how much work customers are willing to hand over.
Encore AI's own customer materials claim improvements including a 30% increase in lead conversion, a 33% conversion-rate increase and, in a savings-account deployment, an 81% lead-conversion increase across 200,000 handled calls. These are Encore AI's case-study figures rather than independently audited results, and the site does not provide enough methodological detail to compare them across deployments.
A narrower thesis in a crowded agent market
The Series A arrives as customer-service and sales automation attract far larger pools of capital. Sierra raised $950 million in May 2026 at a valuation above $15 billion, while Decagon announced a $250 million Series D in January. Attention raised a $30 million Series B in June for AI software aimed at revenue teams.
Product boundaries are also collapsing. Cresta combines autonomous agents, real-time employee assistance and conversation intelligence. Gong has expanded from recording and analyzing sales calls into agentic execution using conversation, CRM and web data. Salesforce, SAP, Zoho and HubSpot already sit on customer records that could support competing products.
Ginzburg's answer is specialization. He argues that historical conversational data has to be the foundation of the agent rather than another input attached to an existing CRM or contact-center stack. That is a defensible technical thesis, though Encore AI will have to prove its mining layer produces better outcomes than competing agents.
The financing backs Ginzburg's effort to move from recommendation software for financial advisers and relationship managers into a broader voice-and-text agent platform.
The opportunity is substantial because every mature sales organization already has a private archive of conversations showing how customers respond. Ginzburg is betting that this archive can become operational software rather than passive training material. The next test is whether Encore AI can turn those company-specific lessons into agents that outperform a written playbook without inheriting every flaw in the conversations they study.