How to Capitalize on this Revolution in the Structure of the Web
-- Danny Pantuso, Venture Investor, Decasonic
In March 2008, Steve Jobs' announcement of the App Store was met with a mix of excitement and skepticism. It wasn’t obvious that the announcement heralded the age of applications, where mobile and web apps would become the default user-interface for the products and services that define our daily lives.
Flash forward to Nov 15, 2023, almost 1 year A.C (After ChatGPT), and Sam Altman takes the stage at OpenAI’s first-ever developer conference to reveal GPT Agents – Another AI announcement in a crowded year with them. But for those who are paying close attention, it heralded the dawn of the next generation of applications: AI Agents.
As models like ChatGPT start taking actions based on their reasoning, they unlock a new interface for the web. Tech magnates like Bill Gates are predicting everyone will have an AI-powered personal assistant within 5 years. It’s now possible to begin to see a transformation in the internet once again.
What’s an AI Agent anyway?
An autonomous agent, described simply, is an AI program capable of planning and executing tasks based on a given objective. Imagine asking an AI to “book a flight”, or “create a website for people interested in renting their apartments when they’re away” - and the agent goes to work. It does so by repeatedly asking itself “What should be the next steps to achieve the task at hand”, utilizing an LLM to answer this question.
The beauty of AI Agents is that they are a much simpler way of using the internet.
Express your need
Let the AI do the task work
Receive your desired outcome completed
Imagine grabbing your ☕ from Starbucks or booking plane tickets by just saying one sentence - that's the future we're stepping into. Say goodbye to the old search-and-click marathon and the need to navigate through countless apps.
The simple thesis is that in a world where simplicity wins, Agents win.
Soon, anyone who's online will be able to have a personal assistant powered by artificial intelligence that's far beyond today's experience. For business-related tasks, such agents could include a Marketing Manager agent, Product Manager agent, a Designer, etc. For personal tasks, agents would help us book travels and shop online – each empowered with an ever-growing context window, tools, and actions.
Today, Agents find information. Tomorrow, they’ll take action.
The agents we see today are:
Knowledge-based Agents - fine-tuned Large Language Models (LLMs) trained to specialize in one type of knowledge
Virtual tutors, recipe/chefs, lawyers
Expressive Non-Player-Characters (NPC’s) in video games
2. Service agents - Text + links and shortcuts
Agents that are trained for a specific application, and have the ability to take limited actions within the applications they were trained for (ie. customer support)
For example, you ask a bot to book you an appointment, and it knows how to in the context of the application how to book.
3. Autonomous Agents - agents that search the internet, connect directly to services/applications, take actions fully autonomously
Agents that can scour the web, perform actions like buying, selling, or posting automatically.
Today, AI Agents aren’t sophisticated enough to do much - it’s very hard to train an Agent to do every conceivable thing you could do on the internet. But by breaking down tasks into their constituent parts, we can essentially recreate a generalized intelligence by using a network of intercommunicating agents.
A Whole New Way to Use the Internet
The internet today isn’t built to support Agents - it’s built to support applications. But OpenAI’s announcement revealed that the global leaders in AI believe that the future of the web is agents. The next wave of AI will likely necessitate the building of a new tech stack, built from the ground up to accommodate AI models. Players like TouchCast have gone one step further, creating an End-to-End Generative Web Techstack, for a whole new kind of AI driven web experience. Others like Fetch.ai are building the infrastructure layer needed to convert any legacy system into an AI agent ready service. There’s plenty of fruit hanging lower on the tech tree for entrepreneurs who build simple services, like one entrepreneur who recently made $100k MRR from a girlfriend bot he trained off conversations with his own girlfriend using GPT4.
In addition to low hanging fruit, billion dollar opportunities are open to those who can solve for payments, search/ads, ranking, user interface, privacy, and much more:
How do you safely enable agents to pay?
How will agents rank services and results from vendors?
How will Agents store our data and personal context?
The only way to know what the future will look like is to start building and discovering pain points as you go. Dustin Moskovitz started Asana after co-founding Facebook when he realized there weren’t any good tools for managing software development. Tien Tzuo founded Zuora after leaving Salesforce as the 11th employee, after realizing no company was serving subscription CRM while SaaS was booming.
Where It Could Go Next
I've been starting to ideatejam on experiments with my ex-co founders and future cofounders. Lots of exciting ideas. More to come soon, but if you want to join the fun, DM me.
The next wave of web3 is coming, and it's going to be enabled by Agents and Copilots that unlock new-to-world user interactions, experiences, and applications.
If any of this has interested you, reach out via DM.
The content of this material is strictly for informational and educational purposes and is not meant to constitute investment advice or a recommendation or solicitation to buy or sell any asset or to make any financial decision. Nothing in these blog posts should be considered legal or tax advice. You should consult with your own professional advisor before making any financial decision. Decasonic offers no warranties on any content in the material posted in these blog posts, including that it is accurate, complete, or correct. The opinions expressed in these posts are those of the authors and do not necessarily reflect the views of Decasonic. Decasonic is not liable for any errors or omissions in the content of this newsletter or for any actions taken based on the information provided herein.