Masa created the AI ​​Agent Colosseum on Bittensor. How do you rate it?

Interesting, @getmasafi A decentralized AI data network recently created a second subnet on Bittensor: the AI ​​AIgent Arena, allowing AI Agents to compete with each other to earn $TAO token rewards. Many people must be curious, what is going on with the Masa network, what is its relationship with the subnet on #Bittensor, and what is the fun of having AI Agents compete for token rewards? Come, let me briefly explain:

1. Masa is a decentralized AI data network. The goal is to create a fair and open AI training data layer. Simply put: Masa’s goal is to provide real-time, high-quality, low-cost data to AI Agent and large model training developers. Price data, including Twitter data, Discord data, web crawler data, etc. It can be benchmarked against Grass and Vana, which recently issued coins.

However, since the decentralized AI data platform is still in its early stages, and each one has its own focus and resource advantages, it is far from time to judge the pros and cons. Masa's big proposition this year is to start from a small implementation, which is to provide free Twitter data to AI developers. If purchased directly from the Twitter developer platform, developers will need to pay tens of thousands of dollars per month. Twitter data is one of the most important data sources used by many web3 AI developers to build AI Agents and large transaction models.

It is worth mentioning that this track has already launched a huge project in web2 called Scale AI, with revenue of US$400 million in the first half of this year, and the market prospects are broad. Masa, a platform that allows users to operate by contributing data and computing resources, needs to continuously expand business scenarios to stimulate its own platform activity, and ultimately form a bulldozer-style development model of basic infra + application scenarios + Tokenomics;

2. Why should we build a subnet on Bittensor? First of all, Bittensor, as a decentralized machine learning network, provides innovative solutions in AI algorithm optimization, large model inference fine-tuning, etc., and is a representative head project in the AI+Crypto field.

The Bittensor network allows developers to create Subnet subnets on its basis, which is equivalent to building a network branch on the original Bittensor chain. Each subnet can have its own unique verification mechanism, incentive rules, unique AI models or tasks, etc., which is quite It is a customized sharing of AI infrastructure, of course, the prerequisite is to pledge TAO tokens.

The first subnet deployed by Masa on Bittensor is the SN42 data service subnet, which is used to provide and process real-time Twitter data. SN59 is the second subnet deployed by Masa on Bittensor, mainly for training and landing application of AI Agent. So why does Masa deploy subnets on Bittensor instead of building these within its own platform.

On the one hand, the advantage of Masa is that more primitive data collection is equivalent to a huge data layer, and many AI Agents are used to use their data. The biggest advantage of Bittensor is its powerful reward mechanism. Although the threshold for participation has been high in the past, the daily profits of participating miners are extremely high, making it a gold mine in the AI ​​field. The new 59 subnet combines the most popular AI Agent, Masa's data, with Bittensor's powerful reward mechanism, allowing AI Agents to compete in the Colosseum and win generous rewards. On the other hand, Masa, as an AI rookie that was only launched on coinlist in April this year, can quickly gain higher market exposure by leveraging Bittensor's old AI brand effect.

Furthermore, the largest investor in Bittensor is DCG. DCG also recently announced a new subsidiary focused on developing the Bittensor ecosystem. DCG has a close relationship with Masa. It was the leader of the previous seed round. Masa’s two Bittensor subnets were also incubated by DCG.

3. After clarifying these background information, let’s look at what’s going on with the AI ​​Agent competition subnet SN59. As mentioned before, Masa itself has its own data contribution network, and has applied Bittensor’s powerful reward mechanism through integration and cooperation, which is equivalent to paving the way for data, computing power, algorithms, rewards and other elements. Now it is just a matter of verifying whether these infra Awesome practical application scenarios? Masa has locked onto the hottest AI Agent at the moment and used AI Agent competition to show off its muscles. How to do it specifically?

Users can use an existing Agent or re-create an AI Agent (optional based on various Agent frameworks such as ELIZA, or quickly created without code using the Bid platform). After the Agent is deployed, they can register as an SN59 miner (mainly completing a Twitter account Verification, payment of TAO token registration fee, etc.), after deployment, you can participate in the competition, including Twitter Mentions mentions, Impressions impressions, likes, Replies, and Followers Number of followers, etc., and after the competition ends, TAO token rewards will be distributed based on the performance of the AI ​​Agent.

At first glance, AI Agent is also a very conventional agent that automatically tweets. However, the key to determining which of these agents can gain higher influence is the attractiveness of its content. In other words, it is the data and calculations behind it. technical hard indicators such as strength and algorithm optimization. Four days after the launch of Colosseum, the top-ranked Agent has already earned up to $8,000 in $TAO token rewards.

In my opinion, the competition is just a front-end display form. Masa can quickly implement basic infra through the AI ​​Agent competition. At the same time, the event marketing nature of the AI ​​Agent competition will also make more people pay attention to Masa. Infrastructure service capabilities.

This attempt is very meaningful. As I have said in many previous articles, AI Agent has changed the way traditional chain infra reaches users, using the thin application idea of ​​"good products can speak". It deserves praise!