Pencil Finance, a decentralized lending protocol incubated by Animoca Brands, says it has completed a $1 million student-loan cycle entirely on-chain, covering capital formation, repayment tracking, and the return of principal and yield to backers. The loans were issued by education-financing company ErudiFi and packaged as an on-chain private-credit investment. The protocol said the funding bundle was backed in July 2025 by Animoca Brands, Open Campus, and NewCampus. According to the companies, the bundle ran on EDU Chain, an education-focused Ethereum-based network built with Arbitrum Orbit, and most of the capital was deployed into student loans and tuition financing in Southeast Asia, including Indonesia and the Philippines. Over a 12-month period, ErudiFi said the capital supported more than 6,600 students across 118 institutions, with about 1,050 students receiving funding directly attributable to Pencil Finance. The structure split investors into senior and junior tranches, with the senior side getting fixed returns and repayment priority, while the junior side took variable returns and first-loss risk. Pencil Finance also said the transaction represents the first fully on-chain student-lending cycle, although that claim could not be independently verified.

AI agent infrastructure platform Agentmuxer has gone live on Base and introduced an open router designed to let AI agents access tools through a single balance. Developed by Amorphic Labs and backed by Y Combinator, the platform removes the need for users to open separate accounts or pay for multiple subscriptions when using services such as web search, data scraping, and image generation. The system is built on the Model Context Protocol, or MCP. According to the source report, it can automatically choose among providers including Exa, Tavily, and Firecrawl, selecting what it determines to be the best option based on quality, cost, success rate, and latency. It also includes an automatic failover mechanism. Amorphic Labs is currently a two-person team. One of its founders, Frank Li, previously worked at OpenAI and NVIDIA. The project is in a waitlist registration phase at the moment.