Consumer-focused artificial intelligence startup PineAI, the business name of 19Pine Pte. Ltd., has launched an agentic AI service designed to handle time-consuming customer service interactions on behalf of users.

The company’s new Pine Voice and Pine Assistant services are aimed at automating what Chief Executive Stanley Wei described as “digital chores,” including negotiating bills, filing complaints, canceling subscriptions and managing disputes with service providers. The company said the agents can conduct phone calls, send emails, navigate websites and complete multistep workflows without requiring continuous human involvement.

While most mainstream AI applications are centered on delivering information, PineAI is targeting operational tasks that typically require lengthy interactions with customer support organizations.

“It’s kind of like a general-purpose AI assistant that helps you deal with digital chores,” Wei said. “It’s very much like a ChatGPT, but it actually does emails, phone calls, faxes and other activities to get your daily jobs done.”

The company said the agents save users an average of 270 minutes and secure roughly $400 in savings through negotiated discounts, refunds and billing adjustments. PineAI cited the example of including a customer who saved $1,900 on automobile insurance and another who reduced a fiber internet bill by $1,800.

The service focuses on scenarios where customer support interactions are often complex or time-consuming, citing telecommunications providers, airlines, insurers, banks and subscription services as common use cases.

Wei said PineAI developed its own voice model rather than relying entirely on traditional speech-to-text pipelines commonly used in conversational AI systems. The company also built its own orchestration framework to manage long-running, multistep interactions.

“Sometimes we need to handle really long tasks,” Wei said. “You get a phone call, you need to analyze transcripts, you need to send a couple of emails and even log into the website and put in your credentials. It’s a long task, and every step relies on the previous step’s success.”

Multimodel

The platform uses models from multiple AI providers, including OpenAI LLC, Anthropic PBC and Google LLC, depending on availability and workload requirements. PineAI’s own models primarily handle voice interactions and task orchestration.

A central challenge is handling unpredictable responses and changing workflows, which require the system to adapt dynamically. Wei said PineAI agents are designed to retry tasks, adjust strategies and draw upon prior successful interactions stored in a knowledge base.

“We built the agent to be resilient,” Wei said. “It does research by itself and also relies on past experience.”

The company must also navigate legal and ethical constraints surrounding AI impersonation and privacy. PineAI’s agents identify themselves as virtual assistants acting on behalf of users during phone interactions. Sensitive personal information is encrypted and processed within what Wei described as a trusted execution environment.

PineAI said it has also implemented anti-abuse mechanisms to prevent harassment or excessive automated outreach. In one case, Wei said, a user instructed the system to repeatedly contact law firms seeking legal representation, prompting the company to impose soft limits on repetitive outreach.

Founded in mid-2024, PineAI recently raised $25 million in funding, which Wei said will be used primarily to expand engineering, marketing and sales operations. The service supports English and Japanese and operates on a subscription basis, with a limited free tier and professional plans starting at $30 per month.

Although PineAI is currently focused on consumers, Wei said business use cases are emerging. Several of the company’s most active users are chief executives at startups who use the platform to manage email, prioritize tasks and negotiate vendor invoices.

The company has also developed an early business-oriented offering that integrates with workplace communications and workflow platforms. Wei said PineAI internally uses the system to monitor invoices, follow up on deliverables and negotiate compensation claims with suppliers and cloud vendors.

“We pretty much run on V2 now,” Wei said, referring to the company’s internal version of the platform. “One of the key qualities of Pine is handling all the back and forth and follow-up.”

Image: SiliconANGLE/DALL-E