Ohad Al-Helou and Uri Cohen didn’t do things by the book. They founded their startup Augmented Intelligence (AUI) seven years ago and have remained hidden until now even though they have raised several tens of millions of dollars from private investors for an AI platform targeting the shopping sector.
Augmented Intelligence now announces the completion of a $10 million funding round at a company valuation of $350 million, bringing the total amount raised to $44 million. This is a big jump since March 2024 when the company raised $22 million at a company valuation of just $130 million.
While generative AI models developed by OpenAI, Google, and Anthropic are capable of producing intelligent answers to complex questions, scanning millions of pieces of information and having an intelligent conversation with the user, their ability to support the decision involving the purchase of a product – a task that is considered the Holy Grail for any company Big technology – non-existent, according to Elhelu and Cohen, who today, completing their funding round, announced a joint distribution contract with Google Cloud.
Neuro-symbolic model
The augmented AI model is not based on the model that has been popular in recent years since OpenAI launched ChapGPT just under two years ago, using a transformer architecture for its engine. The Augmented Intelligence Engine is based on a do-or-don’t environment based on many years of research by founders of decisions made by consumers.
“Language models based on transformers are good at analyzing language and providing information, but they have not proven themselves in decision-making,” Al-Helou told Globes.
Over the years, the company has developed a model known as “neural code,” a neural-like reasoning based on rules that, according to Helo, makes it possible to help support decision-making, for example in a sales field. It’s essentially an intelligent sales agent or service, which is the holy grail for companies like Google, Amazon, and Salesforce.
“If you try to decide which TV to buy or which flight to book based on GPT4 or even Amazon’s own AI assistant, Rufus, you’re going to have a really hard time,” explains Elhelo. Elhelo claims that actions such as comparing product prices according to certain requirements or even booking a flight according to certain criteria are too difficult for existing language models. “How can you trust that an automated agent will book you on a flight that actually lands in New York and that you sit next to the window as you want? The technology on the market doesn’t provide you with reliability in that aspect.”
But the intelligent agent developed by Augmented Intelligence outperforms competing automation engines like Rufus and Salesforce Cloud at these consumer decision-making tasks.
The company declines to disclose details about its distribution agreement with Google Cloud, but industry estimates suggest it will earn AUI between $15 million and $20 million in its first year. Google will not only place the AUI app in its cloud platform app store, but will also market it through a joint sales agreement with potential customers in the retail and tourism industries.
AUI will benefit from a centralized distribution channel, as well as working on the graphics processors in Google AI servers. “While existing transformer-based language models help improve employee productivity, there is a real thirst for action-oriented AI models such as making decisions about online purchases,” says Elhelo. “The need to sell your product inventory is even greater among the big tech giants.”
As far as we know, the agreement with Google Cloud is not exclusive, meaning it allows either company to sign similar agreements with competing companies.
Published by Globes, Israel Business News – en.globes.co.il – on September 30, 2024.
© Copyright Globes Publisher Itonut (1983) Ltd., 2024.