AI Agents- Next frontier to access the web

The history of internet applications has witnessed a continuous evolution, from the era of websites in 2000s to mobile apps in 2010s. However, we are now on the verge of a new phase in web interaction: AI agents. With Gen AI, we have unlocked the ability to access all human knowledge and computers through natural language queries. This will enable widespread availability to services that were once exclusive to the privileged few, including personal assistants, therapists, personal coaches, content creation, coding etc. Most existing consumer applications have will have a AI agent soon, and many new use cases will emerge -> leading to significant increase in productivity and potentially $5T in new market cap.

History of internet applications — websites > mobile > AI agents

2000s was the era when we used websites to interact with the web. Applications were mostly basic and by 2010 ~2B people were online and spent ~1 hour daily. $0.5T of market cap was created. Google, Booking.com and Amazon were the major players.

However, the real transformative power of the internet was unleashed with the advent of smartphones in the 2007. Suddenly, we had the capability of a ‘computer on the go’. By the end of 2020 ~5B people were online, spending ~6 hours daily. Ironically, most incumbents thrived (Google 10x’ed, Amazon 18x’ed, Facebook 10x’ed). But still a lot of ‘new’ value was created by mobile native use cases such as ‘location capability’ (Uber, Dash) or the ‘always on capability’ (WhatsApp, Slack). Almost $4T market cap got created — across incumbents and new businesses.

Now that most people are online and spending a lot of their time online — the next phase of innovation will likely increase the output of the time we spent online. ChatGPT has shown the promise Gen AI will have in massively increasing human productivity. For the first time — we can access all human knowledge via a natural language query. For the first time — all of us — can program / interact / use the computer using natural language. And 200M+ of us are already using it on a regular basis. New use cases have already emerged such as — content creation, writing essays, trip planning etc. Developers using GitHub Copilot are 55% faster. Video editors save 90% time using RunwayML. McKinsey estimates Gen AI could add $2.6T — $4.4T to GDP. Cathie Wood’s estimates are much higher. Either way — this new technology will enable $5T+ of market cap creation in the next 10 years.

Why are Chat Agents likely to be the next frontier?

ChatGPT is one amazing application — but building upon that — we believe that most of our interactions online will move to chat based AI agents. AI agents push the boundaries of generative AI by seamlessly integrating multiple commands/ prompts to autonomously achieve a desired objective, including engaging with the external environment. They achieve this through advanced reasoning and the ability to store information, call APIs or tools, recursively learn, and reason upon themselves. This leaves humans to work on more creative and difficult problems -> which ultimately increases our productivity. No doubt AutoGPT is the fastest growing GitHub repo in history. The dream is to move from 10-blue links, drop down menus, forms on web pages to agents that can work for you based on your preferences and acting as your co-pilot for all possible tasks. Based on our discussions with leading consumer internet companies — we believe that within 2–3 years most apps will have an AI agent.

Where would value be created?

Much like the transition from web to mobile — value will be created across both incumbents and new players. Google, Booking, Expedia, Instacart, Zillow have already announced plans. One under-rated player is WhatsApp which already has distribution advantage (2.5B+ MAUs), and is likely to launch chat based agents. All of this will enable incumbents to further increase penetration, conversion and wallet share — the three holy grails of consumer apps.

More interesting— I believe that new use cases will emerge and that is where zero-to-one value will be created. Some of these new opportunities include:

• Personal Assistant: Bill Gates said at the GS conference last month “Whoever wins the personal agent, that’s the big thing, because you will never go to a search site again, you will never go to a productivity site, you’ll never go to Amazon again.” This includes someone who can book restaurants, doctor appointments, coordinate calendars, shop, book travel etc.
• Digital companion: Chatbots such as Character.ai, Replika, Caryn.ai are leading consumer Gen AI apps. They provide true conversational entertainment and companionship.
• Therapy / coaching: AI powered apps will make therapy, coaching, mentorship much more affordable and scalable. One might even be able to select their favorite celebrity to be their personal coach
• Research Assistant: McKinsey estimates that knowledge workers spend 20% of their time searching for information and chat based research assistants are excellent productivity enhancers.
• Personalized education: Sal Khan says that Gen AI could spark the greatest positive transformation education has ever seen
• Content generation: Apps such as Midjourney, Verb.ai, RunwayML are empowering the masses to generate better art, essays and videos
• Travel companion: Entire arc of travel — discovery, shortlisting, booking, cancellation, change — will be better managed by AI.

Finally, it is important to note that “AI is not the product”. At the end of the day — most of these consumer products are still governed by LTV/ CAC math. Developers who will build engaging, organic, and value added experiences will win. As Kevin Scott mentioned “probably the place where the most interesting products are, are where you’ve made the phase change (due to Gen AI) from impossible to hard. So something that literally you couldn’t do at all before this technology exists has become hard now, because the things that have gone from impossible to easy are probably not interesting”. There will be many “fart apps” and “woo apps” but those won’t be viable businesses. Those will just be fun projects. Someone who solves hard problems using Gen AI will win. And that is the problem statements / founders that we are looking for!