AI Agents need 1 thing for them to 100x from here...
It's something we're all witnessing in real time, AI agents are having their 'DeFi Summer'.
Attention and Capital inflows are matching the sentiment around AI, It's never been a more 'hot' topic in crypto before.
This is the AI supercycle, and these agents are only going to get better from here, I mean that's part of why they're so interesting right, they constantly learn, adapt and improve. Recently, I covered some of the various sub-sectors of these AI agents and why they're so interesting and It really does feel like AI agents can become an integral component of every part of our lives.
Some of my favourites that I've been enjoying trading and speculating on are @aixbt_agent, @AVA_holo, @MycelialOracle, @ai16zdao, @virtuals_io @Thales_ai but also having fun exploring what the likes @TalusNetwork, @OmoProtocol, @Spectral_Labs and others are cooking up.
However, I think we're entering a phase where these agents are good but need improving, they need to be able to do more actions.
Some examples of improvements to AI agents would be:
- Agents that take actionable trades for you based on your style of trading, especially with derivatives. They could even do things like yield farming for you, based on your personal preferences, history and strategies.
- Agents that can create content for you based on your personality, a Waj AI agent for example that would write about all the things I personally love, like DeFi, AI, DePin, @berachain etc.
- Agents that help you even outside of finance and crypto, helping you with your personal admin, your day to day running of tasks, something I would love being on the builder side now too!
In each of the 3 examples, I used a word to describe how these agents need to be - 'personal'.
But here is the problem.
How do you make AI agents personal? Do you willingly just give them your personal data? Do I give the AI agent all my financial history, my health data, my texts and emails, do I just dump my whole brain into an LLM to train it.
I'm not sure I'm ready to do that just yet.
It's why I'm so bullish on @nillionnetwork and their role in the development of AI Agents. The personal data that these LLMs need for training needs encryption in order to ensure our data is safe and secure. The personal data is the way that they'll have the 100x improvements.
Nillion does this through blind computation - fragmenting keys of data to nodes who can compute using privacy enhancing components.
I wrote about how this works here - x.com/0xMughal/statu…
Nillion are already deep in this too. A recently collab with @Meta is already underway building Curl, a privacy based LLM. There's a 31 pager on it if you want to read but perhaps we'll explore that in another future piece!
This is how AI agents are going to make that 100x move from here, they're going to change the world in every way and they'll do it and @nillionnetwork could play a critical role in this.
This piece was written in collaboration with @TailoredWeb3.
It's something we're all witnessing in real time, AI agents are having their 'DeFi Summer'.
Attention and Capital inflows are matching the sentiment around AI, It's never been a more 'hot' topic in crypto before.
This is the AI supercycle, and these agents are only going to get better from here, I mean that's part of why they're so interesting right, they constantly learn, adapt and improve. Recently, I covered some of the various sub-sectors of these AI agents and why they're so interesting and It really does feel like AI agents can become an integral component of every part of our lives.
Some of my favourites that I've been enjoying trading and speculating on are @aixbt_agent, @AVA_holo, @MycelialOracle, @ai16zdao, @virtuals_io @Thales_ai but also having fun exploring what the likes @TalusNetwork, @OmoProtocol, @Spectral_Labs and others are cooking up.
However, I think we're entering a phase where these agents are good but need improving, they need to be able to do more actions.
Some examples of improvements to AI agents would be:
- Agents that take actionable trades for you based on your style of trading, especially with derivatives. They could even do things like yield farming for you, based on your personal preferences, history and strategies.
- Agents that can create content for you based on your personality, a Waj AI agent for example that would write about all the things I personally love, like DeFi, AI, DePin, @berachain etc.
- Agents that help you even outside of finance and crypto, helping you with your personal admin, your day to day running of tasks, something I would love being on the builder side now too!
In each of the 3 examples, I used a word to describe how these agents need to be - 'personal'.
But here is the problem.
How do you make AI agents personal? Do you willingly just give them your personal data? Do I give the AI agent all my financial history, my health data, my texts and emails, do I just dump my whole brain into an LLM to train it.
I'm not sure I'm ready to do that just yet.
It's why I'm so bullish on @nillionnetwork and their role in the development of AI Agents. The personal data that these LLMs need for training needs encryption in order to ensure our data is safe and secure. The personal data is the way that they'll have the 100x improvements.
Nillion does this through blind computation - fragmenting keys of data to nodes who can compute using privacy enhancing components.
I wrote about how this works here - x.com/0xMughal/statu…
Nillion are already deep in this too. A recently collab with @Meta is already underway building Curl, a privacy based LLM. There's a 31 pager on it if you want to read but perhaps we'll explore that in another future piece!
This is how AI agents are going to make that 100x move from here, they're going to change the world in every way and they'll do it and @nillionnetwork could play a critical role in this.
This piece was written in collaboration with @TailoredWeb3.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
