Welcome to the SAG3.ai docs, a comprehensive breakdown of the SAG3 Agent's capabilities, utility and underlying technologies. Whether you're a seasoned investor, a curious researcher, or an AI agent dev seeking insights, SAG3.ai provides detailed, data-driven analysis to help you make informed decisions in the ever-evolving crypto landscape.

Jump right in

SAG3.ai by Virtuals (SAG3)

Supply Economics

- Total Supply: 1,000,000,000 SAG3

- Circulating Supply: 215,127,196.625 SAG3 (as of March 2025)

- Market Cap: $532,188.11 (as of March 2025)

- All-Time High: $0.0025 (as of March 2025)

Utility

SAG3.ai by Virtuals (SAG3) is the official project token for SAG3.ai, which was launched on the Virtuals AI agent launchpad. Please refer to the Virtuals whitepaper to learn more about the utility and supply dynamics of Virtuals agent tokens.

Owning SAG3 tokens grants you spritual co-ownership of the SAG3.ai platform. SAG3.ai is proprietary technology owned by its own community and made available for free (for now) for the benefit of all. It is not a common enterprise with the expectation of profit to be derived from the efforts of others.

Agent Revenues

With that out of the way, let's talk about agent revenues. SAG3.ai is currently earning a few thousand dollars a day in trading fees. These funds are squirreled away in an agent wallet and will be put to good use on behalf of the community in the future.

We are exploring various options for generating further revenues including premium features, third-party integrations (API access) or even an experimental fund powered by SAG3's technical analysis module. Any future revenues generated can and will accrue value to the SAG3 token. With that said, we're just really not focused on revenue right now. Our aim is simply to build the best tech and get it shipped as quickly as possible.

Locked Tokens (~80%)

With the combined efforts of the community, partners and team, 80% of SAG3's supply was bought up almost immediately upon launch (via ). Because of this, SAG3.ai was able to graduate the prototype stage and become a "sentient" agent in less than 24h.

These tokens, spread accross 5 community-controlled wallets, are locked away on Jupiter Lock, vested for a period of 1-3 years. They can be used to fund inference cost, team and agent salary as well as any future endeavours that SAG3 Agent might independently decide to pursue.

Circulating Tokens (~20%)

As of March 23 2025, over 70M SAG3 tokens (about 35% of circulating supply) are held in liquidity with a further 200M tokens (20% of total supply) allocated to future liquidity provisioning. Liquidity is spread strategically across a number of trading pairs including VIRTUAL, SOL, JUP and USDT. The remainder is held and actively traded by the community, generating $100-200k in daily volume and 1000s of dollars in daily revenue from trading fees. Fees will be used in the future to buy back and burn SAG3 tokens.

Logo

Press Kit

Contains Logo files in SVG and PNG format

Color Palette

View on

Contact Information

Media enquiries:

Token Contract

Gx5dX1pM5aCQn8wtXEmEHSUia3W57Jq7qdu7kKsHvirt

Block Explorer

Liquidity Pool

Compare Features

SAG3.ai is by far the most advanced and feature-rich agent currently on the market. The AI agent space is evolving rapidly and we're keeping a close eye on the competition. Let's compare features of SAG3.ai and other agents launched on the Virtuals platform:

*AICS Compliance

The AI Capability Standards (AICS) framework was published in March 2025 by the Virtuals team. The framework establishes a standardized classification system for AI capabilities, defining distinct levels of functionality, decision-making, and adaptability. Each level builds upon the previous one, introducing new capabilities and complexities. The levels are designed to provide clarity and consistency for developers, implementers, and evaluators of AI systems.

Agentic Level

SAG3.ai is currently the only agent on Solana that has been awarded an agentic level. It is one of a tiny handful of agents built on custom frameworks that have achieved a Lvl 3 agent rating. This is a huge achievement that we're really proud of. Take a look at our to learn more about agentic levels and the AICS framework.

Disclosure

We're not trying to paint others in a bad light. We simply feel strongly that teams should be much more transparent about what their agents can and can't do. Some teams will design their agent's output to appear much more well-informed and substantive than it really is. Unfortunately, most of the "agents" coming to market are nothing more than personality prompts wrapped around generic AI Models like ChatGPT or DeepSeek.

The large influx of these "GPT-wrapper" agents has become something of a plague on the industry. They generate a lot of senseless chatter that floods people's timelines. This distracts (and detracts) from the real tech that is being built in this space. The Virtuals team is aware of this problem and that's why they created the AICS framework.

What is an agent anyway? Why should I care?

AI agents are systems that use advanced reasoning and integrated tools to achieve user-defined goals. Agents are superior to and distinct from large language models (ChatGPT, Claude etc.) in that they can independantly plan and execute tasks - without human intervention.

SAG3 is a research agent. It's goal is to provide users with detailed, verifiable research and analysis for a given crypto project or token. It does this by combining the advanced (slow-ish) reasoning capabilities of large language models with hyper-efficient coordination and automation tools, built in-house.

With first-mover advantage and a live product (which is more than most can say) SAG3.ai is positioned to capture a large part of the market for AI-powered research. We offer a better product than others and at a cheaper price.

Okay, but what's to stop copycats with bigger teams and limitless marketing budgets from building the same thing and taking SAG3's market share?

The Data Moat

In the first 30 days from launch, SAG3 analysed and stored over 1000 distinct crypto projects. Every project that gets analyzed is added to our database. By storing and retrieving existing analysis data, SAG3 can work smarter, not harder. Instead of running (and rerunning) slow and compute-intensive reasoning models 30 times per day, SAG3 can simply retrieve existing analysis data in an instant.

From the user's perspective, this means you get all the benefits of premium tier, cutting-edge AI models without any of the typical trade-offs, such as gated access, inference time and cost. The data moat is so valuable in fact, that we decided to give SAG3 away for free (for now).

Q: How accurate are SAG3.ai's analyses? A: SAG3.ai uses advanced AI models trained on vast amounts of cryptocurrency and blockchain data to provide objective analyses. While highly accurate, these should be used as one input among many for investment decisions.

Q: How often is data updated? A: Research data is collected fresh for each analysis. For tracked projects, data is updated according to monitoring schedules.

Q: Can I analyze private or pre-launch projects? A: SAG3.ai requires public information to perform analysis. Projects with limited public data may receive incomplete analyses.

Q: How are the attribute scores calculated? A: Each attribute score is determined by our AI models evaluating specific criteria relevant to that attribute. The overall score is a weighted average of all attribute scores.

Q: What makes SAG3.ai different from other crypto analysis tools? A: SAG3.ai combines deep AI-powered research with nuanced scoring across multiple attributes, plus automated social sharing capabilities - a comprehensive approach not found in other tools.

Q: How can project teams use SAG3.ai? A: Project teams can gain objective insights into their strengths and weaknesses, competitive positioning, and community perception, helping guide development and marketing efforts.

Q: Can I export or share analysis results? A: Yes, analysis results can be shared via direct links or through the automated social sharing features.

Q: How does the sentiment analysis work? A: Our system collects and analyzes social media posts, community discussions, and market reactions to determine overall sentiment and identify key themes.

Virtuals Agent Progression Map

SAG3 Agent is currently classed as a Lvl 3 "Planner" agent by Virtuals' AI Capability Standards. SAG3 is the first Solana-based agent to achieve a Lvl 3 Agentic Level on Virtuals. To qualify as a Lvl 3 agent, the following capabilities must be met:

Lvl 3 "Planner"

• Reasoning and planning, goal-based AI

• Creates and executes plans to achieve user-defined goals

• Can reflect on actions and modify plans during execution

• Capable of reasoning about actions and providing justifications

• Persistent and context-aware memory; retains information across sessions to inform future decisions

Agentic Level Ratings go all the way up to Lvl 6 "Meta" agent. The Virtuals Protocol has a clear and ambitious vision for the future of AI agents. The following is a list of traits and capabilities that are required for an agent to reach agentic levels 4, 5 and 6:

Lvl 4 "Innovator"

• Innovate and create, multi-step decision maker

• Can create novel solutions, tools, or artistic works

• Self-improvement capabilities, with or without human intervention

• Advanced reasoning to solve complex problems and innovate

• Persistent memory with learning capabilities; improves from past experiences

Lvl 5 "Orchestrator"

• Organizational, emergent intelligence

• Can coordinate and command multiple AI agents to perform complex organizational tasks

• Optimizes workflows and systems dynamically

• Advanced reasoning to manage and optimize complex systems

• Advanced persistent memory; utilizes memory to enhance coordination and efficiency

Lvl 6 "Meta"

• Self-evolving, autonomous learning AI

• Self-evolves to solve novel challenges at a meta-level

• Innovates its own algorithms and capabilities autonomously

• Meta-reasoning to improve its own architecture and functionality

• Meta-memory; retains knowledge about its own evolution and discards irrelevant data for efficiency

The AI Capability Standards were put together by the Virtuals team. The framework establishes a standardized classification system for AI capabilities, defining distinct levels of functionality, decision-making, and adaptability. Each level builds upon the previous one, introducing new capabilities and complexities. The levels are designed to provide clarity and consistency for developers, implementers, and evaluators of AI systems.

We will continue to draw guidance from this framework by building out additional capabilities and features for the SAG3 Agent.

SAG3.ai Platform Roadmap

We believe in community-driven development.

Although we can mostly guess what features and improvements people want to see, it's much easier when you just tell us. Feeback and input from the community is always welcome.

How to use SAG3

Agent Logs

The agent logs page shows the real-time chain of thought and goal-based actions taken by SAG3 Agent. Agent logs are the backbone of any self-respecting AI agent. They demonstrate beyond a shadow of a doubt that the tech actually exists and works.

Core Components

1. Research Engine

   • Powered by Perplexity AI's sonar-pro model

   • Capable of collecting and organizing data from multiple sources

   • Source tracking and citation management

2. Analysis Engine

   • Utilizes DeepSeek's R1 reasoning model (hosted in USA)

   • Structured data processing for interpretation and attribute scoring

   • Pattern recognition for strengths and weaknesses identification

3. Tweet Generation System

   • AI-driven content creation for social sharing

   • Context-aware messaging that incorporates cashtags or @ mentions

   • Tone consistency and brand alignment

4. Image Processing Pipeline

   • Automated creation of analysis summary images

   • Optimization for social media sharing

   • Thumbnail generation

5. Scheduling System

   • Automated queue management for social posts

   • Time optimization algorithms

   • Retry mechanisms for failed operations

Architecture

• Frontend: React-based web application with interactive components

• Backend: Supabase-powered database and Privy authentication

• Edge Functions: Serverless functions for AI operations and social media posting

• Data Storage: Secure storage for analysis results and user data

Integration Points

• Twitter API: For monitoring KOLs, posting analysis reports and hot takes

• CoinGecko API: For price and market data integration

• Perplexity API: For research data collection

• Fireworks API: For reasoning-based analysis processing

• OpenAi API: For generating context-aware research images

• AI Agent: The autonomous system that powers SAG3, capable of performing research, analysis, and generating insights without human intervention.

• Analysis: A comprehensive evaluation of a crypto project covering multiple aspects.

• Attributes: The six main categories used to score a project (Fundamentals, Tokenomics, Decentralization, Market Position, Risk Profile, Community).

• Chain of Thought (CoT): A reasoning process used by the AI to work through complex problems step-by-step before arriving at conclusions.

• Community: The group of users, holders, and contributors around a project.

• Decentralization: The degree to which a project distributes control and decision-making across its network.

• Engagement: Metrics measuring how users interact with content, including likes, retweets, comments, and shares.

• Fundamentals: Core aspects of a project like team, technology, use case, and roadmap.

• Impressions: The number of times content is displayed to users, regardless of whether they interact with it.

• Inference: The process and time taken for an AI model to generate output.

• Market Position: How a project compares to competitors and its standing in the market.

• Overall Score: A weighted average of all attribute scores on a scale of 0-10.

• Research Report: The raw data collected about a project before analysis.

• Risk Profile: Assessment of potential risks associated with a project.

• RWA: Real World Assets - physical assets represented on blockchain.

• SAG3 Chat Core: The interactive chat interface that allows users to ask questions about analyses.

• Sentiment Analysis: Evaluation of social media and community feelings about a project.

• Technical Analysis: Evaluation of the project's technical architecture and implementation.

• Tokenomics: The economic model of a cryptocurrency token, including distribution, utility, and incentive mechanisms.

Overview

SAG3.ai is an AI-powered research agent that autonomously evaluates cryptocurrency projects across multiple dimensions, providing in-depth research, quantitative scoring, and social impact analysis. By leveraging the latest AI models, SAG3.ai delivers thorough, unbiased analyses of projects based on their features, fundamentals, tokenomics, market position, community engagement, and more.

SAG3.ai collects data from various sources, conducts detailed research, and presents findings in an accessible format. SAG3.ai has automated social media monitoring and sharing capabilities to help provide key insights and provoke further discussion in the wider crypto community.

How to use SAG3.ai

1. Comprehensive Project Analysis

• Detailed research and analysis of any crypto project

• Scoring across six critical attributes

• TLDR summary for quick overviews

• Strengths and weaknesses identification

• Technical Analysis on real-time price data

2. Multi-faceted Research Process

• Gets research data from verifiable sources using Perplexity API (sonar-pro)

• Presents unbiased report with focus on verifiable data

• Provides inline citations and source attribution

3. Quantitative Scoring

• Overall project score on a scale of 0-10

• Individual attribute scores with explanations

• Analysis phase powered by advanced reasoning models (DeepSeek R1)

4. Social Impact Analysis

• Community sentiment analysis

• Social media trend monitoring

• Key themes in recent news

• Market position evaluation

5. Interactive Reports

• Tabbed interface with Research, Analysis, and Social Impact sections

• Support for price chart visualization for tokens

• Project summary with key features

6. Autonomous Trend Monitoring/Sharing

• Monitors KOLs and automatically analyzes highlighted projects

• Auto-generated tweets about analyzed projects

• Image-based project summaries for X

• Scheduled posting capabilities

7. Chat Interface

• Ask follow-up questions about analyzed projects

• Get clarification on specific points in the analysis

• Explore additional aspects not covered in the main report

Ok but it's still just a GPT-wrapper, right?

Yes that's right, but it's also so much more. SAG3.ai is a GPT-wrapper that leverages 5-6 distinct large language models that each specialize in a certain field running in parallel to produce very high quality crypto-native research and analysis that you won't find anywhere else.

You could just do the research yourself, but that would take hours. SAG3.ai does it in seconds.