Upping the Stakes With AI Agents

Upping the Stakes With AI Agents

Insights from Sophon's COO, @EdC.


We’ve all been exposed to most of the popular applications of AI.

Generative art. Generative music. Hell, even full-on generative videos (most of which seem to be racing to dive headfirst into the uncanny valley).

But really, we’re just scratching the surface when it comes to the ways in which AI can and will be used. The greatest draw of this tech is the ability to use it to automate a lot of the boring stuff we as humans would really prefer not to do (soon we’ll have robots doing laundry and dishes—you’ll see).

One of the more promising selling points of AI lies in AI Agents, which can most easily be understood as digital assistants—think self-driving cars, chatbots and even stuff like the YouTube algorithm that pushes content to you based on your watch history and activity on the platform.

The thing is, AI agents aren’t just tools; they’re dynamic, interactive systems that adapt, learn, and act, making them powerful partners in resolving many of the problems we face today and unlocking new possibilities for the future.


Who’s at the Forefront?

With the huge potential that AI agents hold, it’s little wonder that there’s a bunch of projects scrambling to develop products that the masses can use. Some of the biggest players right now are:

  • Virtuals Protocol - a marketplace on Base that allows users to buy, create and trade AI Agents
  • ai16z - an AI-led hedge fund and open-source AGI movement
  • Spectral Labs - a machine intelligence network designed to generate autonomous onchain agents

Each of these projects is focused on leveraging the power of AI agents and have already made great strides in the space, garnering a lot of interest and expanding their user base as they work to finetune their offerings.

The rise of AI companions

Ever wanted to have a conversation with a superhero? Or maybe you just want to work on your social skills? One of the most popular ways AI Agents are being used at present is in the form of AI companions.

By visiting a sites like character.ai, you can talk to AI versions of characters like Sukuna from Jujutsu Kaisen or even Elon Musk (please don’t mention anything about “let that sink in” though).

Then there’s talkie, which is a great way to start up a whirlwind romance with an AI anime waifu or husbando—if that’s not your bag, you can always talk to Draco Malfoy from Harry Potter or even President Donald Trump. There’s something for everyone here if you care to look hard enough.

Not to be outdone, there’s more than a few web3 projects looking to capitalize on the AI companions craze:

  • Nectar AI - create your own AI girl/boyfriend or choose from a selection of existing ones
  • CharacterX - a decentralized network of AI companions serving multiple needs
  • Oh - working to build the world’s leading uncensored AI model in a responsible way

At face value, these projects seem to be focused on the carnal needs of consumers—but the real goal goes deeper: it’s about companionship. Loneliness is something humans all over struggle with, and AI companions have the potential to address this in a big way.


The Problem

No matter what platform you use or which AI companion you chat with, you can’t help but notice that the interactions are seriously limited—if anything, it just feels like performance theater or roleplaying, and although this may work for some, it fails to grab the attention or interest of a wider audience.

If you’re aware of the illusion, then the experience just isn’t all that great, is it?

The Solution

Sterile text-based chatting and generated images just aren’t gonna cut it. AI agents need to become way more powerful and helpful if we actually want them to have any wider use-case where they’re actually relevant to consumers.

Outside of companions, a really compelling application would be an AI agent that acts as a next-gen yield optimizer where you can give it some instructions on how to invest. For example:

  • You tell the optimizer you want to deposit $10k USDC and optimize for yield on protocols with a TVL greater than $1 billion
  • The optimizer executes, auto compounds and farms according to the parameters you set for it
  • It constantly monitors data from multiple protocols and dynamically rebalances your portfolio while managing risks to maximize your ROI

Back on the AI companion thing, being able to see and hear your companion would make a world of difference. Imagine being able to interact with an AI agent like it was a video or audio call—now that would be a serious gamechanger.

Mercifully, this reality isn’t that far off. Tencent recently showcased its video-to-video AI model that can animate human movements and facial expressions—it isn’t perfect yet, but it’s scarily close to the real thing.

As for AI audio, there’s a ton of projects, but ElevenLabs might be the most renowned—again, not perfect, but we’re well on the way to something truly mind blowing.


A Matter of Time

While AI agents may lack nuance right now, their potential is undeniable. The evolution of these systems is likely to accelerate beyond our expectations and sooner than we think. The projects that succeed in creating genuinely helpful, intuitive digital assistants won't just dominate the market – they'll completely reshape our world.



Disclaimer: This content is provided for informational purposes only, and should not be relied upon as legal, business, investment, financial or tax advice. You should consult your own advisers as to those matters. References to any digital assets and the use of finance-related terminology are for illustrative purposes only, and do not constitute any recommendation for any action or an offer to provide investment, financial or other advisory services. This content may not under any circumstances be relied upon when making a decision to purchase any digital asset referenced herein. The content speaks only as of the date indicated.