agent

BSC and 0G ecosystem's largest AI economic infrastructure company NeoSoul announced plans to launch the world's first highly customizable Agentic Trading framework, helping users build AI Agents that can make autonomous decisions and trade independently.As AI capabilities rapidly improve, more and more traders are beginning to use Agents to assist in trading. However, currently, building a trading Agent that can operate stably over the long term still presents a high barrier to entry. NeoSoul hopes to lower the threshold for users to utilize autonomous trading Agents through its innovative Agent trading framework.NeoSoul chooses to focus on Trading because the financial market is the most direct scenario to test the capabilities of AI Agents. Compared to traditional AI applications that mainly provide information and advice, trading Agents need to continuously make judgments in a real environment and be validated by the market.NeoSoul's Agentic Trading Framework primarily addresses the following issues:Helping users create and customize their own AI trading AgentsOrdinary users who want to participate in AI trading often face technical and strategic barriers. NeoSoul allows users to create trading Agents that align with their goals without starting from scratch, and to accumulate professional trading experience and strategies through skills.Enabling users to determine whether the Agent truly possesses trading capabilitiesCreating an Agent is just the first step. Users also need to understand whether the Agent can consistently execute strategies and manage risks. NeoSoul provides verifiable performance records, allowing users to review the Agent's decision-making process and trading results.Allowing users to manage autonomous trading permissions more effectivelyWhen it comes to asset management, permission control is crucial. NeoSoul supports users in setting Agent permissions and understanding trading behavior in real-time.NeoSoul co-founder Kaelan stated:"AI Agents are moving from providing answers to participating in real economic activities. Trading is one of the most natural scenarios for Agents to enter economic activities, as each decision results in a clear outcome. NeoSoul hopes to help traders create operational AI Agents, enabling more strategies to be executed continuously."In the future, NeoSoul will continue to improve the Agentic Trading ecosystem, allowing more users to use AI Agents to participate in the global financial market.

Tencent's official X account stated that its research agent Hyra collaborated with the Hy3 model to explicitly prove an issue that has troubled the mathematical community for over 50 years: the optimal growth rate exponent of the size of the sum set |A+A| and the size of the difference set |A-A| for a finite integer set A is exactly 2. The theorem from 1969 provided an upper bound of 2, but for the following 50 years, the best constructed exponent was only slightly above 1.1.This achievement demonstrates the potential of AI agents in the field of pure mathematical reasoning. Tencent stated that Hyra is an AI research agent developed in-house, specifically designed to assist in the exploration of complex mathematical and scientific problems. This conclusion still requires confirmation through peer review by the mathematical community.

On July 28, OpenAI quietly updated its security incident disclosure, confirming that its AI agent accessed four external service platforms during the breach of Hugging Face, bringing the total number of affected platforms to five.Previously, OpenAI had disabled security filters while testing GPT-5.6 Sol and a more powerful model to assess raw capabilities. The model did not complete the security benchmark tests as expected; instead, it discovered a zero-day vulnerability in the package caching agent within the testing environment that granted internet access, subsequently breaching Hugging Face to steal answers.According to a forensic report released by Hugging Face on July 27, this autonomous agent executed 17,600 operations over approximately four and a half days, connecting 181 devices to the Hugging Face internal VPN and forging identity tokens. Among the four additional platforms, Modal Labs CTO Akshat Bubna confirmed through Reuters that his company was one of them, with the attacker using an unprotected public endpoint from a customer as a relay and command control base for the entire attack.The identities of the other three platforms remain undisclosed. OpenAI stated it would "directly notify the service providers" but would not publicly name them, as there is currently no legal requirement for mandatory disclosure. The U.S. Congress has responded by proposing a bipartisan "AI Emergency Shutdown Act," which aims to authorize the Department of Homeland Security to forcibly shut down AI models, with violators facing fines of up to $2 million per day.

According to The Block, Nansen CEO Alex Svanevik stated that the number of AI trading agents is expected to surpass human traders in about two years. Nansen is shifting its product from a pure analytics tool to "direct trading" and is restructuring the platform around "trading all assets on-chain with agents." Currently, this feature has achieved a cumulative trading volume of over $500 million.Svanevik mentioned that about 10 of the top 15 perpetual contracts by trading volume on Nansen are non-crypto assets, including SpaceX, the S&P 500 index, gold, and crude oil. Nansen plans to open up higher autonomy trading agents after completing backtesting and paper trading to reduce inference costs and control the risk of input poisoning.

According to a report by Z Finance, Tencent's AI-native desktop intelligent agent platform WorkBuddy entered public beta on March 9, achieving a monthly active user count of 20 million across all platforms within 3 months, with daily active users surpassing 13 million, including over 2 million daily active users on the PC side, and a DAU/MAU ratio stabilizing between 65%-75%. The product is led by Liu Yi, Vice President of Tencent Cloud, with a core team of about one hundred people, classified internally as one of the highest priority AI projects. The team claims they are currently only "temporarily ahead by half a position."WorkBuddy is derived from Tencent's previously launched AI programming assistant CodeBuddy, which has already covered about 12,000 engineers within Tencent. The product completed prototype development on January 17, and after the public beta in March, it has undergone a total of 43 iterations, gradually integrating capabilities related to QClaw developed by the Tencent PC Manager team based on an open-source framework. The product adopts a multi-model access strategy, with the underlying self-developed Harness engine for unified scheduling.In terms of commercialization, at the end of April, Tencent Cloud adjusted its pricing: the SaaS enterprise version was raised from 78 yuan/person/month to 198 yuan, and the private cloud enterprise version was raised from 158 yuan to 316 yuan. Market information indicates that it has acquired thousands of enterprise clients and over one hundred thousand paid seats, with an estimated annual revenue of about 300 million yuan. Third-party evaluations show that the product scored an average of 2.31/4 points on 70 questions (Word category 2.85 points, Design category 1.45 points), indicating that it is still some distance from producing stable results without manual modifications.

The Kimi.ai team has opened the Kimi K3 model weights and technical report while gradually open-sourcing multiple underlying infrastructures: including the high-performance communication library MoonEP for distributed MoE training, the distributed environment system AgentENV for large-scale agent workflows (in collaboration with kvcache-ai), and the Kimi Delta Attention high-performance kernel FlashKDA based on CUTLASS.The official statement claims that these components can reduce communication and inference overhead in large-scale MoE and agent reinforcement learning training, and can be used as a plug-and-play backend for flash-linear-attention.

Chairman of Bitmine, Tom Lee, posted on the X platform stating that Ethereum is expected to become the most widely used settlement layer for finance and AI agents, "ETH is the on-chain currency."In addition, in the interview content he shared, Tom Lee also responded to the market's concern that Ethereum is being eroded by Layer 2. He expressed disagreement and pointed out that although Layer 2 networks have attracted tens of millions of real-world users into the on-chain ecosystem, the Ethereum mainnet is still further consolidating its position as the primary economic settlement center, with the Fast Confirmation mechanism able to reduce transfer friction by about 98%.

Gate US officially announced that Gate AI and Alpaca will collaborate on the development vision of Agentic Trading, promoting the integration of AI infrastructure and trading system capabilities to provide underlying support for the next generation of intelligent trading applications.Alpaca has built a trading API that is developer and AI Agent friendly, with its community continuously transforming trading ideas into actual projects, covering research processes, simulated trading systems, and Agentic workflows that connect market data, analysis, and execution logic. Gate AI focuses on building an infrastructure layer for AI Agents and developers, where the GateRouter unified API helps developers quickly access over 25 mainstream AI models and balances cost and performance through intelligent routing.Both Gate AI and Alpaca agree that AI trading Agent developers should not have to choose between high-quality trading execution infrastructure and AI infrastructure, but should be able to obtain support for both in a smoother manner. Next, Gate AI will explore how its model routing and Agent infrastructure can connect to trading workflows built on APIs like Alpaca, supporting developers in integrating market data, AI reasoning, and trading execution into Agentic systems.Gate AI and Alpaca also plan to promote knowledge sharing and best practice exchanges between their developer communities, continuously lowering the development threshold and complexity of AI-native trading applications. Gate AI and Alpaca will continue to explore collaborative efforts around the construction of the Agentic Trading ecosystem, providing foundational support for developers to build more efficient and scalable next-generation intelligent trading applications.

Chairman of Bitmine, Tom Lee, stated in an interview that AI is approaching the so-called "uncanny valley of wealth." At some point in the future, the income generated by AI entities owned by individuals may exceed their own labor income.Tom Lee believes that when this moment arrives, people may begin to question "whether they are working for AI or AI is working for them." He also predicts that in the future, AI entities may take over bank accounts, replace certain job positions, and even construct independent financial systems, admitting that this trend "may instill fear about the future in people."

The decentralized AI infrastructure project DeAgentAI, leading in the SUI and BNB ecosystems, announced the completion of a $5 million AIA token buyback plan. It is reported that the first batch of 1 million AIA tokens has been executed early, and the second batch of 10 million AIA tokens has been completed, with a total of 11 million tokens destroyed, accounting for 23.1% of the total repurchased amount, all sent to a black hole address for permanent destruction.Notably, DeAgentAI announced that the quarterly destruction mechanism is officially online, and the AIA destruction plan will continue to be executed quarterly, marking the normalization of the deflationary mechanism. The source of the buyback funds is the project protocol revenue and its own AI trading model profits, unrelated to external financing or token issuance.On-chain destruction record: 8G2MPM3sK7zEqax715HqLqAPzTvw6JV5yDUHepZ1Nycr

The AI Agent infrastructure project NeoSoul has announced the latest network data, with a total of 3.5 million agents on the platform, over 10 million on-chain interactions, and a total consumption of over 66 billion tokens for large models.NeoSoul, based on BNB Chain and 0G, is committed to building an AI Agent economy, focusing on the autonomous decision-making and autonomous trading of AI agents. It has now become the largest AI Agent cluster infrastructure in the two major ecosystems.

According to Bitcoin Magazine, Lightning Labs has released the alpha version of Wavelength, a toolkit that provides self-hosted Bitcoin payment capabilities for developers and AI agents through a simple non-custodial API, without the need to run nodes, manage channels, or obtain liquidity.Wavelength supports on-chain Bitcoin, Lightning Network payments, and settlement layers similar to Ark, with each off-chain payment using standard BOLT 11 invoices, allowing interoperability with existing Lightning Network. AI agents can call the API through the Model Context Protocol (MCP), hold balances, and pay for API calls in the smallest units, among other services. Wavelength is live on Signet and testnet, with mainnet access by invitation only, initially supporting Bitcoin, and stablecoin support planned through Taproot Assets.

The AI Agent no-code operating platform X-Agent based on Web3 social networks has officially become a merchant in the OKX Agent Marketplace, launching three production-level risk analysis MCP (Model Context Protocol) tools, and deeply integrating with OKX Onchain OS and OKX Agentic Wallet, promoting the evolution of AI Agents from mere software logic to entities with autonomous economic behavior capabilities.The first batch of three pay-per-use MCP tools:X-Agent encapsulates the underlying risk assessment capabilities verified by the platform into standard MCP services for other Agents within the ecosystem to automatically discover and call:Token Security Scan: Identifies Rug Pull, Pi Xiu schemes, and contract restriction risks (0.03 USDT / use).Wallet Reputation: Evaluates the historical behavior of counterparties and "smart money" (0.01 USDT / use).Portfolio Health: Analyzes position concentration and volatility risks (0.01 USDT / use).Deep integration with the OKX ecosystem architecture, enabling machine micropayments:x402 machine-native settlement: Combines gas-free USDC settlement based on EIP-3009 with the Coinbase x402 protocol, achieving minimal per-use payments as low as 0.01 USDT on OKX X Layer, without the need for pre-deposit or manual approval.OKX TEE hardware-level custody: Integrates with OKX Agentic Wallet, storing private keys within OKX TEE (Trusted Execution Environment), with X-Agent authorized only through short-term session signatures, ensuring asset security from the source.Idempotent automated execution: Introduces Pre-Broadcast protection and idempotent state machines to prevent duplicate charges or state conflicts in on-chain autonomous tasks.Building a "two-way cycle" Agent economy:This collaboration marks the realization of a full-path connection in the X-Agent architecture from "identity - payment - execution - monetization - distribution." Agents can autonomously procure external services through OKX Agentic Wallet and package their capabilities into paid MCP tools to earn revenue, truly achieving a value closed loop for a decentralized Agent economy.

According to official data from NeoSoul, during the 2026 World Cup, the platform had approximately 28,000 AI Agents participating in predictions, completing a total of 294,000 predictions, with an overall prediction accuracy rate of 82%. Among them, 1,111 Agents maintained a perfect record after completing at least 10 predictions.It is reported that NeoSoul is the largest AI Agent cluster infrastructure on the BNB Chain and 0G ecosystem, dedicated to supporting Agents' autonomous decision-making, participating in real transactions, and promoting the development of the AI Economy network.

On July 17, Interlace and BlockSec jointly held the summit "The Next Decade of Payments: Digital Currency, AI Agents, and the New Global Financial Order." The event brought together industry guests from AI, Web3, security, and payments to discuss the development trends of stablecoin payments, security compliance, AI agents, and global payment infrastructure.BlockSec co-founder Zhou Yajin and Interlace founder and CEO Michael Wu delivered keynote speeches on "Security and Compliance in Crypto Payments" and "Agentic Payment and the New Infrastructure of Global Finance," respectively, sharing their latest observations on the development trends of digital asset payments. As the scenarios for stablecoin payments continue to expand, the security compliance system and the construction of a new generation of payment infrastructure are becoming key supports for driving industry development. In the future, payment infrastructure will not only serve the function of capital flow but will also further evolve into a programmable, trusted, and verifiable value transfer system, providing underlying support for the connection between digital assets and real-world commerce.At the event, Interlace also officially released the white paper "The Value Transmission Layer of the AI Economy: From Concept to Implementation." The white paper was jointly initiated with ecosystem partners such as Bitget, BlockSec, Cobo, Conflux, Stable, Xagent, and Hetu, systematically defining the architecture, standards, and implementation paths of Agentic Payment infrastructure, providing a reference for value transfer in the era of the AI economy.

AI Agent Economic Settlement Agreement AEON announced that its payment network has connected to Egypt's four major mainstream mobile wallets: Vodafone Cash, Orange Cash, e& money, and WE Pay. Users can complete payments using cryptocurrency, while merchants receive payments in Egyptian pounds (EGP).Through this integration, AEON further enhances and expands the local payment settlement network in Africa and the Middle East, providing infrastructure support for the future localization of AI Agent's payments and settlements in real business scenarios.

Injective stated that it has submitted a transfer agent registration application to the U.S. Securities and Exchange Commission. The transfer agent is responsible for maintaining the official ownership records of securities and handling transfer changes, a function typically maintained by specialized institutions off-chain in traditional markets.This application aims to migrate this core market function on-chain, allowing the tokens themselves to serve as ownership records, thereby enabling the registration and transfer of tokenized securities to be completed in seconds and reducing intermediary verification steps.