Today, everyone is grappling with how to integrate AI into their business or personal life. One particular field where AI has been popular lately is in social interactions, particularly with AI agents, which have emerged as a predominant use case of generative AI.
Source: a16z
With generative AI, human-to-computer interactions are reaching a new level. From simple rule-based chatbots with no true context of an ongoing conversation, we are evolving to seamless one-on-one conversations. Today, a conversation with a good AI chatbot is nearly indistinguishable from a conversation with a friend. The first logical iteration of AI agents is to have them replace highly repetitive and deterministic tasks or workflows. This includes things like basic data collection conversations, or simple customer service tasks.
But why be so conservative? Given the expected growth of AI models, there’s no doubt that the agents we interact with today will seem like toys in comparison to the depth and complexity of relationships that will be possible. In a future that is not so far off, we believe that AI agents will be fully integrated into our lives. They will become our new personal coaches, therapists, or social companions. It will become the new norm to have multiple domain-expert AI chatbots for each specific need and to chat with them on a regular basis. Want advice on your relationship? Write to your AI love coach. Want to invest in a new startup? Write to your AI VC analyst.
More importantly, we believe that the rise of AI agents will enable new kinds of 'conversations' that didn’t exist before. This could reinvent the form factor of existing services or create entirely new ones. According to McKinsey , AI agents could deliver more than $500 billion in value by 2030. We are on the verge of a Cambrian-like growth phase for AI agents, and the possibilities feel limitless.
Source: alphaplease
From an investor's point of view, we believe that the best way to capitalize on this tremendous market opportunity is to be exposed to fast-growing AI agents platforms that are user-centric, seamless and easy to use, highly customizable, and distribution-focused.
We see Moemate as an interesting project that is building exactly that, and this is why we are excited about this project.
Overview of Moemate
Moemate is a highly customizable AI-driven virtual companion platform that empowers communities to build, share, and monetize AI agents - from character chatbots to intelligent virtual assistants.
Source: Moemate
Currently, the Moemate tech stack equips the AI characters with the following skills:
- Chat: Users can interact with an AI agent seamlessly through either text or voice.
- Image Generation: AI characters can create a wide range of images based on user requests. They can also generate selfies that reflect their unique appearance prompts.
- Web Search: Users can ask AI characters to search the web for the latest information.
- Image Analysis: AI characters can analyze and respond to the content of images uploaded by users.
AI agents creation process
A key feature of Moemate is that it allows users to create and customize their own AI characters.
Source: Moemate
- The basic: First, they have the option to add detailed prompts for their characters, including Bio, Personality, Scenario, Message example, Greetings, First message and System prompt.
- Upload information: Users can upload knowledge to their characters in PDF or TXT formats. A great way of enhancing the character's ability to meet specific needs and stand out.
- Persona settings: Users can also upload detailed information about themselves to the AI characters.This makes conversations much more unique and interesting.
- Train custom LORAs (Low-Rank Adaptation of Large Language Models) for images: Moemate allows users to train their AI on images using a lightweight model training technique called LORA. This feature is ideal for tasks like image recognition.
- Modules: Additional skills can be integrated into existing AI character features through modules. This enables characters to perform tasks such as generating code or playing special games.
AI model
Moemate is leveraging a lot of different premium LLM (Large Language Learnings) and image models, each with particular specializations. Currently it has some of the best models of the industry. This includes Synthia 70B, Nous Capybara 34B, GPT 4 Turbo, Claude V2.1 , Mythalion, Nous Hermes L2 and many more. This multi-model approach enables a broad range of agentic capabilities to better assist the specific need of users.
Source: Moemate
Towards decentralization
Although Moemate was initially built as a Web2 idea, it is progressively incorporating Web3 principles and, in the long run, aims to fully decentralize its stack. From user-owned agents, they are moving to ownable on-chain autonomous agents capable of executing settlements. Furthermore, they are also in the process of implementing their utility token that will be at the center of the Moemate ecosystem as they transition to a decentralized marketplace/network model—a great way to bootstrap supply, accelerate growth, and reward all its different stakeholders.
Source: Moemate
What makes Moemate unique?
Network effect
Moemate has already built an amazing product. They already have an app with some amazing AI bots, and their operations are already profitable. More importantly, the project has already overcome the initial challenge of the beginning, which is how to start attracting enough contributors and users to make the project useful and self-sufficient. As of writing, they have more than 4 million active users and over 1.6 million messages per day, and the stats are growing fast.
Currently, the main ways people engage with Moemate are primarily for fictional roleplay. First, users are imagining a character and talking to them. Secondly, people are engaging with Moemate to make character bots. They’re creating bots, personas, or scenarios in which they have an interest. Thirdly, people also use Moemate to create a domain expert (a history teacher or a loremaster), learn more about the particular domain, and just debate with that character about the specific domain.
Interesting roadmap ahead
While for now Moemate is mainly used for entertainment purposes, the vision of Moemate goes way beyond that. In the near term, the project plans to build a desktop agents version, integrate with VR/AR, and enable its bots on the main messaging platforms like Discord, WhatsApp, or Telegram.
The longer-term objective of Moemate is to become the decentralized AI agent marketplace of reference, where there is a real economy behind AI agents, each trained for specific purposes.
Source: Moemate
This means that Moemate is a bet on an already successful and fully functional Web2 project that is pivoting to Web3 and aims to grow to its full potential by becoming the on-chain AI agent marketplace of reference.
Comparative analysis
The two closest peers to Moemate in both the Web2 and Web3 landscape are Character AI and MyShell AI respectively. When we compare those different projects based on different metrics, we see that Moemate is an interesting one.
Source: Similarweb
