In late December 2022, Google announced a 'Code Red' threat and assigned several staff groups to address the problem. This was not uncommon; through the years, Google has had a lot of its products upstaged by competitors - from social networks to VR glasses. But this time the attack was on Search.
The management declared the company's business model at risk from open source Large Language Models (LLM) like ChatGPT. These LLMs offer greater efficiency savings to users over traditional search, serving up targeted information in easy sentences rather than just piling link after link in front of users. They are able to explain ideas in ways people can easily understand and can even generate ideas and discover solutions - from shopping lists to business strategies.
The actual question here was why a behemoth like Google was raising the alarm for something that it could build overnight given the number of resources it has. The answer was clear: Google makes money from users searching not by finding answers. If Google gave users the perfect answer to each of their queries, no one would click on Google ads - which accounts for at least 80% of their yearly revenue.
For users, Google search has become like a dying mall. As a consumer you may still go there out of habit, but once there, you realize that none of it is what you want and the chances are that you will leave without getting what you came for.
The rise of Large Language Models has exposed Google's flank. They can help users arrive much faster at a solution compared to the traditional search engine route. LLMs are the Amazon.com to Google's dead malls. Yet, there are two big reasons why these models haven't completely replaced the traditional search: firstly they still require the right input and secondly they only give directions to the users - rather than tackling the problems for users themselves.
Throughout history, from Socrates to Google - it has been established that asking the right question in the right context is how one can extract the maximum value out of anything. While every new shiny piece of technology has ways to answer questions better, the ability to ask the right questions remains the core skill and the cornerstone for progress. The same is true for Large Language Models.
ChatGPT - the most widely used LLM in the market - has millions of daily users. Despite that, people need help in supplying the right prompts. ChatGPT is now developing a reputation for answering informational queries and being unsuitable for transactional queries. Internet gurus still suggest search engines as the go-to option when you have to perform an action (or, a transaction) - it could be "book a flight to New York'', or "buy groceries". The engines can then provide users with a list of websites where they can complete their desired action.
The key idea here remains the same: the user still has to take action themselves. The very nature of search implies labor. A majority of the AI tools are focusing on the informational aspect of LLMs - but they are ignoring its biggest advantage: the elimination of user labor.
Large Language Models are poised to replace search with intent. They can offer massive efficiencies to users in completing mundane tasks. We are on the brink of an intent-based economy, which marks a massive paradigm shift away from search to something more useful - to intent-based task delivery.
This is precisely what LLM-driven AI agents do.
From pop culture to scientific discourses, AI usually is referred to as a model that can receive a task from the user, perform action and deliver it successfully. While this is a simplification, companies have been trying to accomplish it since the last decade - only to discover that this process could not be achieved by Large Language Models alone. It needs a set of three items:
1. LLM that can build the right context around the user's intent.
2. Conversational technology that gathers the right details.
3. Entities that can execute a user's request.
While there have been several works in place around (1) and (2), it has been extremely difficult to execute all three of them as a set.
Until now.
Over the last decade, Fetch.ai has been busy building the tech stack to enable an intent-based task delivery model into existence. This year we have decided to come out of R&D and release the technology into your hands.
We're proud to announce the DeltaV platform: an intent engine that uses chat to match user queries in real time to relevant apps and services listed on an open network for business.
Remember the last time you wanted to take a family vacation? Remember having to open fifteen tabs on your browser to search for the right places? The time you had to spend on Tripadvisor, booking.com, Skyscanner, and ten other travel blogs. The time you had to spend on budgeting, finding the best prices, planning your itinerary, booking - and managing all of them, and more.
Now imagine that you had a personal assistant to take care of all of that. Your own Alfred - who just needs to know your preferences in order to organize, plan and book various aspects of your vacation: from flights and accommodations to local dining spots. He could also help you with personalized activities on the vacation - and keep you informed about any real-time updates regarding changes to travel plans, like flight delays or cancellations.
Becoming that butler is just one of the infinite possibilities that Fetch.ai's platform can accomplish.
Fetch.ai introduces a new chat interface called DeltaV. Here's how it works:
The brilliance of the new DeltaV platform lies in its ability to abstract away all the complexities for the user. It not only undertakes all intent-based tasks but it can also follow through with concrete action, for example: booking services and ordering goods. It utilizes Fetch.ai's unique peer-to-peer network to connect with any business or service.
The Fetch.ai platform is made up of two major components:
DeltaV: The Large Language Model Task Builder
The Agent Network: The Framework for Open Marketplace.
The platform enables a new economy: open dynamic marketplaces. Any organization could jump on the network easily to get more users and increase their revenue. It eliminates the hassle for newer companies to get discovered and puts the quality of the product and user experience above all.
It is market meritocracy at its finest.
Do you represent a business or a service? Join the Agentverse and get discovered by users. Get early access via our waitlist on Fetch.ai's website.