Google, Meta Palona Ai, is released, bringing emotional clients to non -technical institutions

Speaking of myself, the interaction with Chatbot in which artificial intelligence works in any trader on his website is a frustration exercise. The phone trees are usually worse. I was betting that I was barely alone in my evaluation. Among us did not experience long times of elections, slow responses, lack of updated information and awareness of the customer’s account date, and granting exciting politeness and a set of other shortcomings?

New emerging is called Pallo For the first time last week, this aims to fix this unfortunate situation. It provides direct institutions for the consumer-thinking about pizza stores and electronics sellers-with customer support sales agents around the clock throughout the week, which uniquely reflect the personality of the brand for each company, sound, inventory stocks, and providing value. The electronics seller contains a “processor” made by Balona, ​​while the pizza store gets the character of the elegant agent in Server – naturally.

In all cases, Palona focuses on creating artificial intelligence agents with “eq” or “emotional intelligence / emotional arrangement”, building them from a group of open source Amnesty International models and training some human sociology research.

“A type of basic thesis that the company is that we can create something delightly and feel real, like a real human conversation,” said co -founder of technology (CTO). Tim Hosh, In a personal interview with Venturebeat recently. “Chatgpt is a very useful tool, but it does not feel a human conversation.”

Palona claims that its system can be easily implemented through a non-technical brand on its website, mobile phone application or phone lines-with uniquely designed responses to each brand and each communication environment. In fact, its agents are already working on orders to deal with work, answer questions and complaints, suggest products and raise customers.

Strong foundation background

In addition to hows, Palona is participating in its founding and led by a team of engineers from some of the best technology companies in the world, including: Maria ChangPalona CEO, is a former head of engineering in Google and VP/GM of AI products in Meta and CTO of Tinder. It was also founded both, which Yahoo gained in 2013.

Senior scientist in Balona Steve Leo, PhDAnd he was previously scientists at the Samsung AI and Tinder Center. LIU is also an IEEE fellow and Canadian Academy of Engineering, with more than 390 research papers for his name.

And hows itself is the participant inventor of the online data storage system, and is the co -founder of LoudCloud and OpsWare (the last HP obtained for $ 1.65 billion). He was also the chief technology official in Netscape, HP Software, and produced the Meta Ai developer.

“We are building completely independent sales agents – not tools for sales delegates, but they are actually actual sales delegates,” Zhang said in the same direct interview, adding that it was the “employee of the century.”

24/7 polite, distinguished sales agents, elegant

Palona Ai places itself as a solution for companies looking to improve their sales performance, customer participation and brand loyalty.

Palona agents work as dedicated employees for custom sales, and combine soft sales skills with 24/7 available, unlimited capacity, calling advanced memory, and they can interact with customers with Chatbot format via the Internet, or text number/text that can be compatible with them, or even the sounds that Works on behalf of the phone.

“100 % – we support the sound,” Zhang explained, “for example, in the arrangement of pizza, the sound is still a major style for the user. In the Middle West, about 50 % of people are still connected to it. On the eastern and western coasts, about 20 % But it is still important. “

Palona voices are licensed, but the company has the ability to train and publish custom voices, and even audio cloning for accredited or CEO representatives, let’s say.

The company has realized, through a test, that the audio version of AI sales agents in Palona will need to have a clear different reaction patterns of the text.

“We have experienced different audio reactions, and in the pizza order, for example, customers wanted competence,” Zhang. “They didn’t want Amnesty International – they just wanted to complete their orders as quickly as possible. So we have improved this, making it a lower personality, less treatment, and more efficiency.”

Unlike traditional chat groups that work as assistants for human representatives, Palona AI is designed to deal with entire sales courses without human intervention.

“There is a big gap between vibrant artificial intelligence models such as ChatGPT and what companies already need – Amnesty International’s agent can sell, transfer and complete UPSell.”

Balona claims to reduce errors and reduces hallucinations from artificial intelligence by up to 98 %, ensuring reliable reactions.

Zhang and hows said that even for the most analog companies, it was a short team to go with Palona, ​​just several days for a simple implementation.

Hawaz said that the Foundation provides Balona with “common questions and evidence of employee training, policies and procedures.”

After that, they determine with the Palona what actions the agent must take – whether it is to treat orders, respond to inquiries, or deal with support problems.

“The largest factors that affect the preparation time are: the amount of integration required with the current systems of the customer (for example, pos, CRM, demand platforms). If we already support their system, it is delivery and operation.” “If it is a system we already support, the agent may be ready within two days. If they are using a new unfamiliar system, it requires additional engineering work, which may take longer.”

In addition, Zhang said that “in fact in the process of preparing the agent’s automation. In the end, companies will be able to use the Palona agent to form its Palona agent!”

Three linguistic models are better than one

Palona is done with all this by combining three different models: The first is a large, dedicated language model (LLM) that works as a basis for each distinguished business sales agent-The pizza store gets a different tone and personality from the electronics seller, and each one is allocated to each customer outside the box.

There is also a supervisory model that discovers and gets hallucinations and removes it from the main model before taking it out to the customer.

Finally, the system also includes a small memory tracking model (SLM), allowing it to build deep profiles for customers based on previous reactions-eliminating personal conversations and stronger customer relationships.

Hawis said: “Artificial intelligence is very good in remembering everything at all, which leads to a terrible experience,” due to the introduction of unimportant details. “Or does not remember anything, which, again, is a terrible experience,” because she will not know what happened previously in the conversation.

To overcome this, Palona trained its small model to help LLM to manage its memory.

This is the “task of this small model” to know, “Well, what is important here, what is not important”, in every conversation, Hawr said.

In general, the company included an interesting data collection for all its models that reflect a wide range of human emotional responses that all sales and humans face customers (and agents) to deal with them regularly.

“We have trained our models to get the high emotional (emotional intelligence) by wiping literature in psychology, and determining what makes someone intelligent emotional,” Hows told me. “It turns out that there is eight dimensions of EQ, and we trained artificial intelligence on these standards.”

Palona has also trained its models to perform “gentle persuasion”, and respond to customers who have a sense of humor, expressive symbols, and allergies.

Early customer testimonies: Wise, Mindro, and my heart pizza

Palona AI already works with many brands for consumers, including Wyze, Mindzero and Pizza My Heart, to enhance customer reactions and increase sales remittances.

Wyze, a smart home camera company, has combined artificial intelligence sales aides working in Palona to customize customer support.

Yun Zhang, CEO of Wyze, said in a statement: “I always wished to be able to communicate personally with every customer to share the reason for their choice of the camera and the camera. I have made Paluna that possible with Wyze Wizard AI, where she delivered the customized and personal purchase trip that I wish To give her every customer myself.

Mindzero, a global studio specializing in treatment in contrast, is used, Palona AI agents to answer customer questions naturally and strictly. David Simierad, CEO of Mindzero, highlighted the effect of technology: “Palona, ​​Jen’s agent has become our best friend – able to answer customer questions in a real, nice and useful way, it truly blows our minds.”

Pizza My Heart, a well -known West Coast Pizza series, has transformed her brand character into an active interactive experience. Palona AI’s agent allows the company, “Jimmy The Surfer”, to customers to seek pizza through sound or text while engaging in great conversations.

Chuck Hamers, CEO of my heart pizza company, commented on the experience: “I brought AI Jimmy to life in a way that makes me smile – which I did not expect from artificial intelligence. Now, Jimmy helps my customer seek pizza by talking to them directly – VIA VOICE or Text .

Three columns of Balona success

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

1. The brand consistency – artificial intelligence agents are trained on the company’s knowledge base to ensure that the conversations remain correct with the brand’s voice and identity.

2. EQ high (emotional intelligence) – Palona customers adapt to conversations and outperform both human actors and artificial intelligence models competing in emotional participation.

3. Persuasion-Artificial intelligence determines the main sales opportunities, and proactively recommends to increase sales in the right moments to increase revenues and the value of the agent (LTV).

I have seen all these things at work in a short illustration offer for the Chatbot Customer Sales “processor”, even when I complained about this from my dog ​​that barks on my neighbors (sorry, guys!), I managed to redirect the conversation smoothly to a home camera that I can buy to help Monitor them.

What is the following for Palona?

“We went intentionally after the most difficult problem first – cases and conversion,” Zhang pointed out. “It takes more technical development, but if you can solve it, expansion in other areas such as customer support and loyalty is easier.”

As Amnesty International continues to form the future of trade, Balona approach indicates that customer relationships should not be sacrificed for automation – instead, can enhance artificial intelligence, customization, leadership transfers and construction of permanent brand loyalty.