PineAI launches "Execution AI," responsible for customer service calls and refund negotiations

Consumer-facing AI startup PineAI launches an agent-based AI service that can handle “digital chores” such as customer service calls, subscription cancellations, refund requests, and fee negotiations. Its core capability is not only providing information but also executing phone calls, emails, and website operations in practice.

PineAI is the trade name of 19Pine Pte. Ltd., which recently announced the launch of the “Pine Voice” and “Pine Assistant” services. These services represent users in interactions with telecom companies, airlines, insurance providers, banks, and subscription services, focusing on automating complex and time-consuming customer support tasks.

CEO Stanley Way describes it as a general AI assistant for handling “everyday digital chores.” Way states, “It may look like ChatGPT, but in reality, it can send emails, make calls, send faxes, and so on, until the job is done.”

According to PineAI, users save an average of 270 minutes, and through discount negotiations, refunds, and bill adjustments, they save an average of $400 (about 579.4k KRW). Examples include saving $1,900 on car insurance premiums (about 2.75 million KRW) and $1,800 on fiber internet costs (about 2.61 million KRW).

The service views lengthy and variable customer service processes as primary application scenarios. For example, tasks such as requesting to lower communication costs, resolving airline disputes, or disputing insurance claims—requiring multiple confirmations and follow-ups. PineAI explains that in these tasks, the AI can analyze call content, send follow-up emails, and log into websites when necessary, handling multiple steps continuously.

Technologically, it has built its own speech model and orchestration framework. The company states that existing conversational AI mainly relies on converting speech to text before processing, but PineAI designed an independent architecture to manage long-duration, complex tasks. Since the success of each step depends on the previous one, it requires higher resilience and process management capabilities than simple reactive AI.

At the same time, it utilizes models from OpenAI, Anthropic, Google, and others to handle unexpected responses.

The PineAI platform dynamically selects from multiple AI models based on task conditions and availability. Its proprietary models mainly handle voice interactions and task coordination. This structure is designed to cope with the unpredictable reactions and frequent dialogue changes typical in customer service environments.

Way says that PineAI’s agents are designed to retry, adjust strategies when a task fails, and leverage a knowledge base of past successful cases. He explains that the system operates in a “resilient” manner by self-investigating and referencing past experiences.

However, such services also raise legal and ethical issues. PineAI states that during calls, the agent will inform the other party that it is a virtual assistant acting on behalf of the user. Additionally, sensitive personal information is encrypted and processed within a secure area the company calls a “Trusted Execution Environment.”

The company has also established mechanisms to prevent abuse. Way mentions that a user once set the system to repeatedly contact multiple law firms seeking legal representation, prompting the company to introduce a “soft limit” on repetitive automated contacts. This is clearly a measure recognizing that while AI automation increases convenience, it could also be misused for excessive contact or harassment.

Secured $25 million in funding… enterprise-level expansion possibilities emerge

Founded in mid-2024, PineAI recently secured $25 million (about 36.2 billion KRW). The company plans to mainly invest this in expanding engineering, marketing, and sales teams. Currently, the service supports English and Japanese, with a limited free plan and paid plans starting at $30 per month (about 43.46k KRW).

Although initially focused on consumer services, enterprise needs are beginning to surface. According to Way, some of the most active users are startup CEOs who use the platform for email management, prioritizing work, and negotiating bills with partners.

PineAI has also developed early enterprise products integrated with internal communication and business platforms. The company states that internally, they are using the system for bill monitoring, delivery time confirmation, and negotiations with vendors and cloud providers. Way reveals that the company’s own operations largely depend on an internal version of the platform.

This release demonstrates a trend in the generative AI market shifting from “question-answering AI” to “task-executing AI.” While PineAI’s customer support automation can significantly boost productivity, issues such as identity disclosure, personal data protection, and setting boundaries for automatic contact are likely key conditions for the service’s wider adoption. Ultimately, the competition among agent-based AIs will depend not only on performance but also on “trust” and “controllability.”