This morning, Yzi Labs announced its investment in Tensorplex Labs, immediately triggering a 500% surge in its Bittensor subnet project, Dojo. In the same month last year, the decentralized AI startup Tensorplex Labs completed a $3 million seed funding round. One year later, Tensorplex Labs has contributed to the Bittensor ecosystem across various aspects, including DEX, data dashboards, and AI. This article will provide a brief analysis of how the two-year-old Singapore-based company has become a key participant in the Bittensor ecosystem.

The DeFi Core of Bittensor

Staking and Lending

Before the launch of dTao, the Bittensor ecosystem was relatively underdeveloped. Tensorplex contributed core DeFi products to the TAO ecosystem, one of which is Tensorplex Stake. As a cross-chain staking product, it introduced stTAO, an LSD (Liquid Staking Derivative) ERC-20 token. This allows EVM investors to participate in the TAO ecosystem and earn staking rewards without needing to download a Substrate-compatible wallet or manually delegate. Additionally, staked assets can be bridged to networks like Bittensor to participate in validator operations. To facilitate this, Tensorplex established a minimized-trust bridge between Substrate and EVM. Beyond staking, it also incubated lending protocols like TaoBank, enabling users to use staked TAO as collateral for interest-free loans. The total value locked (TVL) in staking peaked at over $30 million, making Tensorplex one of the largest validators on TAO.

Exchange + data dashboard integration backprop

Just last month, as dTAO was officially launched, @BackpropFinance also went live. At this point in time, as one of the three DEXs where dTAO can be traded—alongside HotkeySwap and TaoStats—Backprop not only allows users to trade various Alpha subnets but also provides a comprehensive data dashboard. It integrates trading dashboards, address analysis, on-chain information queries, market analysis, and other features into a single product. With a familiar UI and a well-developed data dashboard, it significantly lowers the entry barrier for new users joining the TAO ecosystem. This has made Backprop the most popular DEX product within Bittensor.

AI Ecosystem Development

LLM

Tensorplex has made multiple attempts in large models, creating a model that outperformed all open-source counterparts in the decentralized LLM pre-training competition on @MacrocosmosAI’s “Subnet 19.” Operating validators across more than 60 Bittensor subnets and AI chains such as @flock_io and @AlloraNetwork, Tensorplex has been at the forefront of innovation. Since 2023, it has been deploying top-tier models on Bittensor Subnet 9, including the release of Sumo. In April 2024, after increasing the parameter limit to 7 billion, the Sumo-T9-7B model even outperformed well-known models like Falcon 7B and Llama 2 7B.

Tensorplex Stream

Tensorplex has also explored the field of AI Agents. In collaboration with @Corcel_X’s “Subnet 18,” it built the first Web3 RAG chatbot, Tensorplex Stream. This chatbot gathers information from cryptocurrency podcast transcripts to train an AI Agent, providing the latest cryptocurrency summaries and insights. Beyond the chatbot, Tensorplex has also developed Podcasts, enabling AI to “speak” the information it collects. Leveraging RAG’s advantages, this approach to gathering real-time insights in the fast-evolving crypto market is now being widely adopted.

Dojo Subnet

In the past 24 hours, Dojo Subnet has become the most prominent subnet within dTAO, surging 500% after YziLabs announced its investment in Tensorplex Labs. Dojo is a data collection platform—but more than that, it is an integrated ecosystem where humans act as third-party observers. Human intelligence can directly influence AI development, training, and agent operations both on-chain and in the real world.

The core workflow allows anyone to generate unique synthetic tasks using LLMs—such as game physics feedback, image generation quality, or physics algorithms. Dojo collects human feedback data and applies response obfuscation techniques. Miners then distribute tasks, while human validators verify the basic facts generated by LLMs, assessing the quality of responses from contributors. Rewards are allocated based on multi-dimensional metrics: human consistency (weighted Cohen’s Kappa), response stability (intraclass correlation coefficient ICC), and factual fidelity (Spearman correlation). This data is used to improve and optimize open-source models.

Dojo also implements technical measures to prevent Sybil attacks and ensure contributors are real humans. As the system matures, additional mechanisms will be introduced to deter attackers, forming a robust contribution chain. To date, Dojo has collected over 3 million human-generated data points.

Yzi Labs’ investment in Tensorplex Labs is just a reflection of the growing market recognition of the Bittensor ecosystem. Often overlooked in mainstream discussions, Bittensor is home to long-standing AI projects with significant expertise and dedication, fostering an environment of serious building. While issues like superficial projects or an evolving dTAO framework still exist within subnets, there is optimism that Bittensor will continue delivering high-quality products. Rhythm BlockBeats will keep a close watch on its progress.

Disclaimer：

1. This article is reproduced from [BlockBeats]. The copyright belongs to the original author [BUBBLE]. If you have any objection to the reprint, please contact Gate Learn team, the team will handle it as soon as possible according to relevant procedures.
2. Disclaimer: The views and opinions expressed in this article represent only the author’s personal views and do not constitute any investment advice.
3. Other language versions of the article are translated by the Gate Learn team and are not mentioned in Gate.io, the translated article may not be reproduced, distributed or plagiarized.