From: thepipeline_xyz
Monad Madness in Bangkok showcased a diverse array of projects leveraging artificial intelligence (AI) and deep technology, spanning decentralized data platforms, advanced computational models, and novel applications in various industries [00:00:48]. These ventures highlight the future growth and potential of new technologies when combined with blockchain.
Advancing AI Scalability and Inference
42
42 is a protocol dedicated to addressing the scalability of AI, a challenge that traditionally limits the number of API requests to centralized providers [00:25:08]. The team, with over a decade of experience in AI projects, including self-learning AI agents and conversational AI platforms, is building a “planetary layer for AI scalability” [00:24:52], [00:26:04].
Their approach leverages:
- Small Language Models (SLMs): Noting rapid progress, such as Qwen 2.5 which can run on a MacBook Air and surpass GPT-4 in certain tasks like coding [00:26:17].
- Excess Compute: Utilizing the largely untapped processing power of modern consumer devices, where users typically use only about 20% of their computer’s capacity [00:26:47].
- Swarm Inference: A self-supervised inference mechanism where nodes collaborate, peer-reviewing responses and compiling a knowledge tree to generate accurate answers [00:27:24]. This method significantly lowers inference costs (by 100x) and improves AI accuracy by mitigating issues like hallucination [00:28:22].
42’s research paper on these findings was presented at MIT’s Decentralized AI Summit, and they are preparing to launch their devnet [00:28:52].
Decentralized Data and Specialized AI Applications
Ever Network
Ever Network is building a decentralized healthcare data storage protocol on Monad [00:17:13]. They aim to solve the problem of siloed and centralized medical records by giving users ownership and control over their health data [00:16:08]. This includes features like an AI-powered chatbot that allows users to inquire about symptoms without needing a doctor’s visit [00:17:45]. The platform also enables users to monetize their data by licensing it to pharmaceutical companies, insurance providers, and drug researchers, or by being recruited for clinical trials [00:18:14]. Ever Network has established government contracts, notably with the Thai government, managing medical records for over 650 hospitals and 10 million patients [00:18:39].
JoJo World
JoJo World is a decentralized AI 3D data platform focused on collecting high-quality 3D data for training large world models and AI 3D models [01:01:46]. They incentivize creators, particularly UI/UX professionals and architects, to contribute 3D data using tokens [01:02:41], [01:06:05]. This data, used by companies like Google and Nvidia, is crucial for developing spatial intelligence for robotics and virtual simulations [01:01:53], [01:02:04]. JoJo World has accumulated over 1 million raw spatial 3D datasets and is working with partners such as Nvidia and Microsoft [01:04:10].
Primus
Primus specializes in cryptography technology, offering a compute and verification network that combines Multi-Party Computation (MPC), Transport Layer Security (TLS), and Zero-Knowledge Proofs (ZK) with Fully Homomorphic Encryption (FHE) [02:21:20]. Their aim is to build a cryptographic layer for secure data intelligence, connecting authenticated, high-value data from Web2 to Web3 and AI applications in a trustless manner [02:23:03]. Primus claims their protocol is 10 times faster for TLS verification and 300 times faster for verifiable computation compared to other solutions [02:24:00]. They offer data verification capabilities for digital goods marketplaces, verifiable data markets, and prediction markets, with a long-term goal of deploying FHE for confidential AI [02:24:47].
Score
Score Technologies is bringing “Moneyball” to the world of sports, starting with football (soccer) [02:58:07]. Their decentralized network of football intelligence uses over 200 AI models that compete to provide insights from game data [02:59:21]. These models analyze vast amounts of data—up to a million data points per match [03:00:34]—to help coaches with tactical shifts, discover new talent, prevent injuries, and even create automated media content [02:59:51]. Score plans to open-source their models and ingest 200,000 games per week [03:00:48]. They have partnerships with football figures like Harry Kane [03:01:46].
Totem
Totem is the first AI deepin network designed to unlock the value of daily conversations [03:59:58]. They provide a device that attaches to a phone, capturing voice data and converting it into real-time transcription, translation, and summarization [04:11:11]. This data is then encrypted using ZK proofs to become anonymized datasets, which users can upload on-chain via Monad’s fast throughput in exchange for tokens [04:12:26]. The hardware is designed by an Apple expert, and the AI models are optimized by experts from Google and Alibaba [04:41:47]. Totem aims to monetize through device sales, premium subscriptions, and selling anonymized data to AI companies for model training [04:42:59].
Blockchain’s Economic Layer for AI
G
G is building the economic layer for AI and compute, enabling the creation of new types of on-chain assets backed by “real AI cash” [03:39:02], [03:42:42]. They aim to provide direct exposure to the GPU asset class, which they argue can generate high yields (100-200% APL) [03:40:07]. G tokenizes GPUs and their yield, offering DeFi use cases like GPU-backed stablecoins, lending, borrowing, and structured products [03:45:48]. Their architecture is designed for transparency and decentralization, with an orchestration layer for continuous monitoring of underlying GPU assets [03:47:08]. The yield is generated from demand for AI compute, with one co-founder’s existing AI cloud company (GMI Cloud) already generating significant revenue from AI cloud services [03:48:47].
Other AI/Deep Tech Engagements
- Dusted: While primarily a “token-centric chat” platform [00:06:30], Dusted was mentioned in the initial list of categories that included AI and DeFi [00:00:48].
- Talentum: This platform for user acquisition in Web3 unifies Web3 and Web2 data credentials to target skilled users in the creator economy [03:32:17]. They plan to integrate an AI agent to manage campaigns from start to finish [03:35:28].
- Paper Plane: This lifestyle app focuses on rewarding merchants and consumers for uploading consumer data on-chain [02:12:55]. In their long-term vision, they aim to gather authorized on-chain data to help build AI agents for the future [02:16:45].
- Moo Digital: This RWA protocol brings institutional-grade corporate lending on-chain, offering yields from public and private credit deals [02:03:32]. Their competitive advantage lies in their expertise in asset selection and credit underwriting, leveraging their traditional finance backgrounds [02:06:08], with the goal of bringing new and differentiated inventory to the Monad ecosystem [02:06:25].
The presence of these projects underscores the significant intersections between AI and blockchain and the continued push towards innovative decentralized solutions across diverse sectors.