The field of#AIis opening a new era for humanity, it is gradually creeping into work and daily life. Besides, the development of#blockchainhelps create a world that operates automatically based on code and a#cryptoeconomy.
Although there are still many challenges, the idea of ââcombining these two leading fields is of increasing interest. This combination is further promoted after a series of AI-labeled crypto projects have had strong growth in the past period.
An important part of the AI ââBlockchain system is the infrastructure, it needs to be operated on a platform that meets the speed with abundant data feed. In the article below, we will learn about a project in this field called Spectral @Spectral_Labs.
What is Spectral?
#Spectral's story begins with the project arguing that today's AI is mostly a "black box" operated by large entities in the market. The user sends a request and then receives an answer back from the black box without knowing what it did, what factors the result was created from, or if there is anything shady or unusual in the answer. .
By combining efforts with the blockchain system, Spectral aims to solve this problem.
Spectral @Spectral_Labs is an infrastructure project for a system that combines AI and Blockchain. Through Spectral, users can create and use open source AI products in an efficient and transparent manner.
Products and operating models
#Spectral has two main products: Spectral Syntax and Spectral Nova. Spectral Syntax is a toolkit that helps users create onchan AI products, or in technical terms, Agents, while Spectral Nova acts as a network that provides computational data and connects Agents.
We will learn in turn in the sections below, but first let's clarify the concept of Agent that the project refers to.
» Agent Economy
The term AI Agent was mentioned after the launch of ChatGPT @OpenAI in November 2022, after which it was widely used in both the technology and Web3 fields.
AI Agent in the context of Spectral is a set of on-chain instructions and code that can execute automatically.
Specifically, some distinguishing points of Spectral AI Agent are as follows:
âą Is a set of instructions that can be executed consistently.
âą Designed to operate autonomously in the on-chain space based on real-time data and events.
âą Have access to its private key wallet to perform transactions without the user's presence.
In short, you can understand that AI Agent is a complete product including on-chain smart contracts and related systems created by users to serve a specific purpose. These AI Agents, in addition to solving problems, can also help their creators earn income by allowing other users to use them.
» Spectral Syntax
Spectral Syntax is a toolkit that allows users to create their own AI Agents.
Through a series of Large Language Models (LLM) optimized for writing Solidity code, users can easily create different AI Agents just by entering the requirements into the chat box.
Spectral Syntax includes many modules that work together to operate, including:
âą Base Models: Basic data processing and communication system.
âą Agent Identities: Agent identification system to differentiate and evaluate effectiveness.
âą Agent Knowledge Bases: Data feed for Agents.
âą Plugins: Used for retrieving other necessary external data such as oracle feeds, data from the internet, ...
âą ML Inferences: Machine Learning (ML) system for computation.
âą Wallet Management: Wallet management system. Spectral uses the Account Abstraction (AA) wallet according to the ERC-4337 standard.
Spectral Syntax's operating process is as follows:
âą Step 1: First, at the user interface, the user enters the requests that need to be processed.
âą Step 2: Orchestrator will analyze and build context for the request through the LLM system.
âą Step 3: Based on the processed requests, the Orchestrator will search for suitable Agents and then transfer the request to the Agent.
âą Step 4: Agent Handler then queries the additional data needed to build an advanced request set with enough information to prepare for the response generation process.
âą Step 5: Agent Handler calls the Language Model (LLM) and sends the advanced request set to calculate the response. Then send it back to Orchestrator to determine if additional plugins need to be called to get additional data if necessary. This process is repeated until the Orchestrator thinks the result has met the standard.
âą Step 6: Finally, the results are sent back to the user through the user interface to complete the process.
» Spectral Nova
Spectral Nova is a machine intelligence network that serves as a decentralized data feed. Spectral Nova encourages researchers and engineers in the field of Machine Learning to build data provision models for Web3 applications.
Main ingredients in Spectral Nova:
âą Creator: The person who issues the challenges, the standards of the challenges, along with the rewards for the winning solver. Creator will receive a portion of revenue from end users.
âą Solver: The person who solves the challenges given by the Creator and has the opportunity to receive rewards from the Creator. They then also receive a portion of the revenue from usage by end users.
âą Validator: Ensures that models are processing work with integrity and accuracy. Validator checks Solver's response through zkML Proof.
âą Consumer: Search and use suitable services created by Creator and pay fees.
Spectral Nova is designed to be modular with many components, which allows it to increase scalability and flexibility. They include: Integrated wallet, Blockchain (Arbitrum), Account Abstraction, RPC node (Alchemy), Pulse (on-chain data tracker), Modeler CLI (interaction management), Nova (on-chain federation and off-chain), Verifiable computation (proof calculation), IPFS. I will not go deeper into the technical details to avoid complexity. You can find out the details at the project's Whitepaper page.
Development roadmap
Spectral Nova launched and launched the first Challenge in November 2023 with the participation of more than 400 Solvers and about 20% of models received prizes from the challenge.
Spectral's 2024 and 2025 development roadmap includes the following stages:
âą Q1/2024 - Inception: Launch Machine Intelligence Network and launch SPEC governance token on Alpha Mainnet version.
âą Q2/2024 - Scaling: Launch the second Challenge, the Solidity LLM system. The Validator system is decentralized and attracts Creators.
âą Q3/2024 - Diversification: Diversification of inference feeds such as NFT Recommendation Engine, Uniswap X Filters, Multimodal Solidity LLMs.
âą Q4/2024 - InferChain: Deploying on-chain AI/ML Inferences on testnet.
âą 2025: InferChain Mainnet.
Development team
The project does not disclose details about the development team, below is some information I collected through the project's LinkedIn page.
âą Sishir Varghese - CEO & Co-Founder: He started with Spectral in June 2020, with significant experience in technology and finance. Join Kernel Fellowship at Gitcoin for 3 months, Strategic Partner at Loopring for 2 years and 7 months. Before that, worked as an Architect at The Jerde Partnership for 2 years and 2 months.
âą Mihir Kulkarni - Head of Product: He has a lot of experience in the blockchain field. Before becoming Head Product at Spectral, he was SVP at Truist, building a decentralized identity service. At Coinbase, he improved the technical support process. He also developed the Ethereum contract platform at EY and designed blockchain tokens at Sacred Capital.
Investors
Spectral has gone through 3 rounds of capital raising with a total amount of $29.82 million raised.
âą The first round took place in November 2021, raising 6.75 million USD led by Polychain Capital.
âą The second round took place in August 2022, raising 23 million USD led by Social Capital, General Capital. In addition, this round also has the participation of many famous funds in the crypto world such as Circle, Jump Capital, Franklin Templeton.
âą The latest round is IEO on the Gate Startup platform with a capital raised of 70,000 USD.
Tokenomics
" Basic information
âą Token token: Spectral
âą Ticker: SPEC
âą Maximum total supply: 100,000,000 (100M)
âą Network: Ethereum
âą Standard: ERC20
âą Contract address: 0xAdF7C35560035944e805D98fF17d58CDe2449389
âą Status: TGE and listed
» Allocation
âą Investors: 38%, locked for 12 months then unlocked linearly for 18 months.
âą Team & Contributors: 22%, locked for 12 months then unlocked linearly for 30 months.
âą Airdrop + MM: 10.3%, Fully unlocked at TGE.
âą Foundation Treasury: 29.7%, no information on unlocking schedule yet.
" Uses
âą Is a payment tool for network activities.
âą Staking at Spectral Syntax or Nova to enjoy privileges such as becoming a Validator, reducing service fees, and unlocking other benefits.
âą Participate in protocol governance activities.
Epilogue
AI is a big trend not only in the technology field but also in humanity. The combination of#blockchainand#AIis always the desire of technology lovers in both fields. While many challenges remain, recent efforts are pushing us closer to this ideal.
Spectral @Spectral_Labs has a mission not only to connect blockchain with AI but also to create an open economic framework and automated operations for decentralized AI applications. Currently the project is launching the first problems for this vision, we will continue to monitor developments to make more accurate decisions.
Above is all the information about Spectral - infrastructure for blockchain and AI, hope the article helps you in your research process.