Former Google, Meta Leaders launches Palona AI, bringing custom, emotional client agents in non-technological enterprises

Speaking of yourself, interacting with the chatbot, powered by AI on their website, is often an exercise in powerlessness. Telephone trees with robot voices are usually worse. I would bet that I was hardly alone in my assessment. Which of us has not experienced many times, slow answers, lack of updated information and awareness of the customer’s own account history, the provision of courtesy of arts and many other ineffectiveness?

A new startup called Paralo He made his debut last week, which aims to correct this regretful state of affairs. It supplies businesses directly to the users-think about pizza shops and suppliers of electronics-with alive, 24/7 agents for the sale of customers who are uniquely reflecting the personality of the brand of each business, voice, inventory stocks and value proposal. The electronics seller has a “magician” agent made by the Palon, while the pizza store receives a personality of the dude surfing agent – naturally.

In all cases, Palona focuses on creating AI agents that have a high EQ or “emotional intelligence / emotional coefficient”, building them from a combination of open source and own AI models and training of some of their own, using a human study Sociology.

“A kind of main thesis that the company is that we can create something experience that is delightful and feels true, like a real human conversation,” said the co -founder and chief technology director of the Palona (CTO) Tim Howe, In a personal interview with Venturebeat recently. “Chatgpt is an extremely useful tool but does not feel like a human conversation.”

Palona claims that its system can easily be implemented by a brand that is not technologically on their website, mobile applications or telephone lines-with answers, uniquely tailored to any brand and any communication environment. In fact, his agents are already at work processing orders, answering questions and complaints and offering products and costs to customers.

Fellow

In addition to the Howuz, the Palon is a co-founder and managed by a team of engineers from some of the best technology companies in the world, including: Maria GianPalona CEO is a former Vice President of Google, AI VP/GM for Meta and Tinder CTO products. It also founded the same acquired by Yahoo in 2013.

Mantler Steve Liu, DoctorA former chief scientist at Samsung Ai Center and Tinder. Professor from McGill at McGil University, Liu is also an associate of IEEE and the Canadian Academy of Engineering, with over 390 research documents in his name.

And Howes himself is the co-inventor of the industrial standard system for access to open source directories (LDAP) online data storage system as well as co-founder of LoudCloud and Opsware (the last Acquired by HP for $ 1.65 billion). Earlier, he was Chief Technology Officer at NetScape, HP Software and led the performance of developers at Meta Ai Infrastructure.

“We are building completely autonomous sales agents – not tools for sellers, but actual AI traders,” Gian said in the same live interview, adding that this was the “employee of the century.”

24/7 polite, distinct, custom sales agents

Palona AI is positioned as a solution for companies that want to improve their sales, customer engagement and brand loyalty.

Palona agents act as personalized virtual sales employees, combining soft sales skills with 24/7 availability, unlimited capacity and advanced memory recall and can interact with customers in the online chat format, SMS/text number with which they can meet Or even voices powered by AI by phone.

“100% – we keep a voice,” Gian explained, for example, when ordering pizza, the voice is still a major consumer model. In the Midwest, about 50% of people still call to order. It is about 20%on the east and west coasts, but it is still significant. “

The votes of the Palona are licensed, but the company has the opportunity to train and unfold a personalized, even voice cloning of authorized customer representatives or CEO, say.

The company realized by testing that the voice version of Palona’s AI sales agents would have to have clearly different styles of interaction from the text chatbot.

“We have tested different voice interactions and for ordering pizza, for example, customers want effectiveness” related to Gian. “They didn’t want chat AI – they just wanted to place their order as quickly as possible. So we optimized for it, making it a smaller person, less multidarily, more efficiency. “

Unlike traditional chatbots that serve as helpers to human representatives, Palona AI is designed to deal with whole sales cycles without human intervention.

“There is a big gap between realistic AI models such as Chatgpt and what the AI ​​agent actually needs, which can completely sell, convert and UPPELL,” Gian explained.

The Palon claims to minimize errors and reduce AI hallucinations by up to 98%, guaranteeing reliable interactions.

Gian and Howe said that even for the most analog businesses, it’s a short leading team to go with a palm, just a few days for a simple performance.

Enterprise provides Palona “FAQS, guidance for training employees, policies and procedures,” Howe said.

They then define with a palm what action the agent should take – whether processing orders, answering inquiries or processing support problems.

“The largest factors affecting the setting time are: how much integration is needed with existing customer systems (eg POS, CRM, order platforms). If we are already supporting their system, it is inclusion and play, “House explained. “If this is a system we already support, the agent can be ready in a few days. If they use a new, unknown system that requires additional engineering work that can take longer. “

In addition, Gian said the Palona was “actually in the process of automating the agent’s settings. In the end, businesses will be able to use a palon agent to configure their own agent on the Palona! “

Three language models are better than one

Palona does all this by combining three different models: the first is a custom, a fine-tuned large language model (LLM), which serves as the basis for each individual agent for sales of the business-shop for pizza receives a different tone and personality from the supplier of electronics And each of them is customized for every customer outside the box.

There is also a supervisory model that detects, captures and removes hallucinations from the main model before bringing them to the client.

Finally, the system also includes a real-time memory tracking model (SLM), which allows it to build deep customer profiles based on previous interactions-the water on personalized conversations and stronger customer relationships.

“AI is very good at remembering absolutely everything that leads to a terrible experience,” House said, because he brings out minor details. “Or remember nothing that is a terrible experience again,” because he won’t know what happened before in the conversation.

To circumvent this, Palona trains her small model to help her Base LLM manage his memory.

This is “the work of this small model to understand:” Okay, what is important here, what is not important, “in every conversation, House said.

In general, the company included an interesting set of data for all its models, which reflects the wide range of human emotional answers, with which all sales and people targeted at the client (and agents) have to handle regularly.

“We have trained our models to have a higher EQ (emotional intelligence) by exploring literature in psychology and identifying what someone actually intelligent is doing,” Howe told me. “It turns out that there is an oostem definition of EQ and we were training AI against these indicators.”

The Palona also trained its models for the implementation of “gentle persuasion” and responds to clients with humor, emoji and sensitivity.

Early customer reviews: Wyze, Mindzero and Pizza My Heart

Palona AI already works with several user brands, including Wyze, Mindzero and Pizza My Heart, to improve customer interactions and increase sales conversion.

Wyze, a company for smart home cameras, integrates AI Palona sales assistants to customize customer support.

Yun Gian, CEO of Wyze, said in a statement: “I always wanted to contact every client personally to share why they should choose our camera plan and Camplus. Palona did this possible with the Wizard Ai Wizard agent, delivering the curse, personalized trip to buy, I would like to give every customer alone. “

Mindzero, a wellness studio specializing in contrast therapy, uses Palona AI agents to answer customer questions naturally and engaging. David Semerad, CEO of Mindzero, emphasized the impact of the technology: “Our agent of the Palona, ​​Jen, has become our new most good friend – she is able to answer clients’ questions about such an authentic, kind and useful way, He really blows our mind. “

Pizza My Heart, a well -known pizza chain on the west coast, has made its brand a person an interactive AI experience. Palona AI agent of Jimmy The Surfer allows customers to order pizza through voice or text while engaging in light conversations.

Chuck Hammers, CEO of Pizza My Heart, commented on the experience: “Palona Ai comes to life in a way that just makes me smile – I never expected from AI. Now Jimmy helps my customers order pizza by telling them directly – via voice or text. “

Three pillars of the success of the Palona

AI Palona’s solutions are built on three basic principles:

1. Brand coherence – AI agents are trained in the company’s knowledge to ensure that the conversations remain true to the voice and identity of the brand.

2. High EQ (emotional intelligence) – Palona agents are dynamically adapted to conversations and superior both human representatives and competitive AI models in emotional engagement.

3. Conviction-AI identifies key sales opportunities by proactively recommending exceeding sales and cross sales at the right times to increase the revenue and value of customer life (LTV).

I saw all this in action in a brief demonstration of a chatting agent for sales for clients of “magician”, in which even when I complained about my dogs barking on their neighbors (sorry boys!), He managed to redirect the conversation to A smoothly A home camera that I could buy to help follow them.

What is next for the Palon?

“We were deliberately left first after the most difficult problem – shed and transformation,” said Jan. “This requires more technical improvement, but if you can allow this, expanding to other areas such as customer support and loyalty programs is easier.”

As AI continues to form the future of trade, the Palon’s approach suggests that customer relationships should not be sacrificed for automation – at the moment, AI can improve personalization, drive realization and build a permanent loyalty of the brand.