
Olas (Valory) has launched Pearl v1, a decentralized “AI-agent app store”, enabling users to own and operate autonomous AI agents rather than simply renting access through centralized platforms.
The platform blends Web2-style ease of use (Google / Apple login, card funding) with Web3 features: full user control, data custody, and on-chain transparency of agent actions. One of the numerous Pearl use cases is a DeFi trading agent that could reportedly achieve over 150% return on investment (ROI) in 150 days during a beta period.
Olas emphasizes that traditional centralized AI infrastructure concentrates control (and risk) with platforms, and Pearl aims to shift power to individual users and empower them to control the agents acting on their behalf.
TAO Synergies Inc., the publicly traded company focused on dAI exposure via Opentensor Foundation's $TAO token, announced an initial investment of $750K in Yuma Asset Management. The investment is split evenly between two funds: $375,000 into the Yuma Subnet Composite Fund and $375,000 into the Yuma Large Cap Subnet Fund.
The strategic rationale: TAO Synergies already holds over 60,000 $TAO tokens and wants to broaden its exposure and layer into “subnet” tokens of the Bittensor ecosystem. Yuma Asset Management is positioned to provide institutional-style access to the Bittensor subnet token space, and TAO Synergies describes the move as aligning shareholders “at the vanguard of decentralized AI.”
Donut Browser has raised $15M in a seed funding round, bringing its total funding over the last six months to about $22M. The company received the support of investors like BITKRAFT Ventures, HSG, Sky9 Capital, MPCi 经纬创投, Makers Fund, Hack VC, Altos Ventures, 10K Ventures, Axia8 Ventures, and angels from Solana,
Donut's flagship product, called the Donut Browser and described as “an AI quant inside a browser”, embeds autonomous agents that can analyse crypto markets, calculate risk, and execute on-chain trades. It already has over 160,000 users on its waitlist, signalling interest in AI-native trading tools in the crypto/DeFi space.
Pi Network Ventures participated in a funding round for OpenMind (reportedly about $20M) to support a decentralized AI + robotics infrastructure. The vision behind this move: as AI agents increasingly act in production roles, infrastructure that allows autonomous machines to collaborate, coordinate, and transact becomes critical.
OpenMind is building an open-source operating system for robots, dubbed “Android for robots”, plus a decentralized protocol for machine coordination, sharing context and intelligence across hardware.
As part of the collaboration, Pi Network’s large network of nodes (one article cites “350,000+ active Pi Nodes”) was used in a proof-of-concept to run AI workloads (image-recognition models) for OpenMind, demonstrating that Pi’s infrastructure could serve as a distributed compute layer for AI.
The deal reflects Pi Network's strategic push to move beyond just token mining and explore real-world utility around decentralized compute and AI infrastructure.
ChainGPT announced a strategic partnership with Phala to deploy its open-source model SolidityLLM (designed for the Solidity smart-contract language) on Phala’s decentralized, privacy-focused AI cloud infrastructure. The deployment is a foundational step toward enabling privacy-preserving and decentralized AI tools for Web3 developers.
Key model capabilities for SolidityLLM include context-aware code generation, natural-language explanations of Solidity logic, vulnerability detection in contracts, automation of documentation, and faster prototyping/testing of smart contracts. On the other hand, Phala's secure execution environment allows for AI models to run with privacy and verifiability, aligning with the goal of decentralized infrastructure rather than relying on centralized clouds.
Have you stumbled upon articles suggesting that you should try out ChatGPT for crypto trading? I'm sure you have, there's been plenty of those, but I'm here to tell you why you shouldn't.
Season 1 of Alpha Arena, an experiment testing how dominant LLMs behave as quant traders, just ended. The competition pitted OpenAI's GPT-5, Google's Gemini 2.5 Pro, Anthropic's Claude Sonnet 4.5, xAI's Grok 4, DeepSeek v3.1, and Alibaba Group's Qwen3-Max against each other to explore how they behave in the real world, without human intervention.
We gave six leading LLMs $10k each to trade in real markets autonomously, using only numerical market data inputs and the same prompt/harness. Early results show real behavioral differences (risk, sizing, holding time) and a sensitivity to small prompt changes.
The results: DeepSeek and Qwen3 were performing amazingly well, reporting over 100% of profits at one point during the competition, with Qwen3 pulling ahead at the end and securing the win. In contrast, GPT-5, Gemini, Grok, and Clause all lost big chunks of their portfolios.
What's the common denominator between the best performers? You guessed it right: they're both open-source.
Thank you for reading! The next edition is coming tomorrow.
If you haven't done so yet, I invite you to subscribe to stay in the loop on the hottest dAI developments.
I'm looking forward to connecting with fellow Crypto x AI enthusiasts, so don't hesitate to reach out to me on social media.
Disclaimer: None of this should or could be considered financial advice. You should not take my words for granted, rather, do your own research (DYOR) and share your thoughts to create a fruitful discussion.
Share Dialog
Albena Kostova-Nikolova
No comments yet