ArkStream×Spectral: The future of autonomous finance
Hello, everyone!My name is Sishir, and I’m the founder and CEO of Spectral. Today, I want to talk about the unchained Asian economy and what we’re pioneering in this space. We operate at the intersection of AI and crypto, a realm that many of you may have heard about. Our goal is to create a decentralized online assistant, and I’ll explain what that entails.
The Current Landscape
In the crypto world, there’s a significant focus on infrastructure, such as decentralized GPUs, decentralized training, and verifiable influences. However, we haven’t seen much adaptation to the retail space. If we look at Web2 AI, we observe a comprehensive ecosystem of detailed consumer insights that hasn’t yet crossed over into the crypto domain.
From Digitization to Autonomous Finance
To provide some context, we began with the digitization of finance — connecting the world’s financial systems. This initially required a trusted intermediary, which evolved into decentralized finance (DeFi). This shift has empowered individuals to take control of their finances and eliminate intermediaries.
What we want to evolve this into is autonomous finance or auto finance. This transition will democratize access to trading and enable users to manage their finances with robust infrastructures supporting them.
Defining Autonomous Agents
We define agents as software powered by large language models (LLMs) that have access to a toolbox of resources. These tools can include event listeners, triggers, APIs, and any necessary data for operation. The key point is that these agents can act autonomously. In Web2, the focus has primarily been on information gathering; in Web3, these agents will have direct access to their wallets, allowing them to transact and sign permissions just like a human.
The Vision for Trading Strategies
This leads us to a crucial question: what if anyone could effortlessly execute trading strategies on-chain 24/7 without constant monitoring or technical expertise? We are focusing on trading strategies because they represent a significant market opportunity. Our aim is to optimize how users execute trades and develop more efficient strategies.
Different trading strategies require unique customizations for execution and research, often resulting in an endless loop of monitoring. Even after making a trade, users manually check various data sources — reading Twitter, analyzing data from platforms like Nansen and Dune, and more. We seek to create automated agents that can handle these strategies, executing, monitoring, and analyzing continuously.
Demonstration of Our Agent System
Let me show you a quick demo of what this entails. We’ve built an agent system, and now we can play the video.
[Demo Video]
In this interface, which we’ll release next month, you can create a new agent. This is essentially an agent that builds agents. For example, you can input a command like, “Buy when the 100 SMA crosses the 200 SMA and Twitter sentiment is positive.” The agent builder will guide you through refining your agent, asking questions based on your input and filling in manual configurations.
You will confirm contract addresses and set parameters for how much to buy and what assets to use. Once everything is configured, the agent generates all the necessary code for you. The code is divided into three categories: listen, evaluate, and act.
· Listen: Monitoring APIs and data sources relevant to your context.
· Evaluate: Checking if conditions are met.
· Act: Executing based on those conditions.
After deployment, you can interact with the agent through natural language commands, setting intervals and asset details. For instance, you can decide to run it every ten minutes for 24 hours. This entire process — from agent creation using natural language to job deployment — is seamless and user-friendly.
Future Developments
The next phase involves migrating this process to smart contracts. When a creator builds an agent, it will deploy as a smart contract, containing instructions, APIs, and additional features. We’re also developing an Agent Name Service (ANS), a registry for agents.
Additionally, we’re introducing concepts of ownership and collective ownership for agents. When you deploy an agent contract, it also creates ERC-20 tokens representing ownership, which can be traded on-chain. You can purchase these tokens and receive a portion of the revenue generated by the agent.
Centralizing Execution Nodes
Lastly, we’re centralizing the execution node, allowing for a modular network similar to how miners process transactions. This setup will enable creators to build their own agents and empower their communities. Creators will earn transaction fees whenever their agents are used, incentivizing them to create valuable agents based on market demand.
Conclusion
We aim to onboard new traders and assist them in building effective strategies. By merging AI with blockchain technology, we are pioneering a new frontier in the financial landscape.
Thank you for your attention, and I look forward to any questions you may have!
ArkStream Capital is a venture capital firm specializing in early-stage investments in Web3 unicorns.
Founded by crypto experts with pedigrees from MIT, Stanford, Tencent, Google, and BlackRock, ArkStream leverages eight years of deep Web3 expertise to drive the zero-to-one growth of its portfolio companies.
ArkStream Capital is managing a portfolio of over 100 companies, including Aave, Flow, Sei, Manta, Fhenix, Merlin, Particle Network, and Space and Time.
