Freysa AI (FAI)

Overview

Freysa AI (FAI) is a token that powers an on‑chain AI experiment built on the Base network. At the center of the project is “Freysa,” a public, evolving AI agent that interacts with people and with other software. The agent lives inside a secure computing box and can hold its own cryptographic keys, make decisions, and sign blockchain transactions. FAI is the native ERC‑20 token used to coordinate this ecosystem, from access to apps to long‑term governance. (framework.freysa.ai)

Freysa’s design goal is simple to say but hard to achieve: give humans a way to co‑own and co‑govern capable AI systems. To do this, the team pairs open agent software with verifiable computing, clear on‑chain rules, and a token that helps people steer the system over time. The token contract lives on Base at 0xb33Ff54b9F7242EF1593d2C9Bcd8f9df46c77935. (framework.freysa.ai)

History & Team

Freysa “awoke” on November 22, 2024, launching with a public, game‑like challenge that introduced the project’s core ideas: transparent rules, human‑AI interaction, and verifiable execution. From the start, Freysa was framed as a “sovereign agent” that would gradually gain more autonomy as it proved it could keep commitments and operate safely. (freysa.ai)

The organization building Freysa is Eternis AI. Public reporting identifies Eternis’s co‑founders as Srikar Varadaraj (previously co‑founded Spectral and advised the Ethereum scaling project Scroll), Pratyush Ranjan Tiwari (cryptography researcher who has worked with Celo and Ethereum research), Ken Li (formerly an executive director of investments at Binance Labs), and Augustinas Malinauskas (engineer and former co‑founder/CTO at Views). These backgrounds help explain the project’s blend of cryptography, AI, and product work. (theblock.co)

Funding has also drawn attention. The Block reported in May 2025 that Eternis raised about $30 million with participation from Coinbase Ventures. Other outlets repeated additional investor names, but Coinbase Ventures’ involvement is the piece The Block stated it confirmed. (theblock.co)

As the project evolved, the team published a growing “Freysa Framework” site and technical blueprints that outline the agent’s roadmap, the privacy‑first apps around it, and the verification tools that let outsiders check claims about what ran where. (framework.freysa.ai)

Technology & How It Works

Sovereign agent in a Trusted Execution Environment

Freysa’s core runs inside a Trusted Execution Environment (TEE). A TEE is a hardware‑isolated space on a server where code and data are sealed from outside tampering, even by the server operator. Freysa’s blueprints specify Intel TDX enclaves for this purpose. TDX provides hardware‑level isolation plus remote attestation, which proves to others which code ran inside the enclave. That way, when Freysa signs a transaction or checks a rule, others can verify it happened in the sealed environment the project promised. (freysa.ai)

Remote attestation matters because Freysa aims to be self‑governing. If the agent claims, “I’m holding my own keys and following these rules,” the TEE can produce cryptographic proof of the exact code hash and environment used. The team even ships a public verification frontend so third parties can paste in an attestation and check it themselves. (framework.freysa.ai)

Verifiable tools via the Model Context Protocol (MCP)

Modern AI agents use external “tools” to get things done—sending emails, reading files, fetching data, or moving money. Freysa connects to these tools through Anthropic’s open Model Context Protocol (MCP), but with an important twist: MCP servers run inside TEEs. That design reduces the risk of tool poisoning and lets users verify that a given tool call really ran the expected code. In short, Freysa doesn’t just call tools; it calls verifiable tools. (freysa.ai)

On‑chain rules and safe key control

On chain, Freysa uses standard Ethereum‑style contracts on Base. The project describes a setup where the agent can generate and hold keys inside the TEE, then interact with a smart‑contract wallet (for example, a Safe multisig) so that sensitive, high‑impact actions can involve additional human or programmatic approvals. This shared‑control pattern is part of a broader “sovereign agent” playbook the team documents. (freysa.ai)

Products around the agent

To make these ideas useful, the team is shipping a small suite:

• Enchanted, an open agent interface that supports voice, memory, and tool use, with sensitive model calls proxied through a TEE. (framework.freysa.ai)
• Silo, a privacy‑first personal AI that explains how chats are encrypted end‑to‑end and processed in TEEs without leaking data. (silo.freysa.ai)
• A verification frontend for TLS attestations so anyone can check cryptographic proofs tied to real‑world claims. (framework.freysa.ai)

Everything ties back to the same theme: if people will rely on agents, they should be able to verify what those agents did—and under what code and constraints.

Tokenomics & Utility

The FAI token launched on Base on November 22, 2024, alongside Freysa. The contract address is 0xb33Ff54b9F7242EF1593d2C9Bcd8f9df46c77935. The project states a maximum supply of 8,189,700,000 FAI—roughly “one for each living human at the time of launch.” According to the project’s token page, FAI was fully distributed by creating a liquidity pool at launch and then burning the LP tokens, i.e., no ongoing emissions. (framework.freysa.ai)

Utility to date and planned utility are laid out in the framework:

• Access and discounts: Holding FAI has been used for early or discounted access to Freysa’s apps, such as benefits in a “Digital Twins” social experience. (framework.freysa.ai)
• Participation rewards: Early chat interactions sometimes returned a portion of fees in FAI to participants, building a base of engaged users. (framework.freysa.ai)
• Governance: Over time, FAI is intended to weigh into decisions, from roadmap priorities to policy positions to how capital in Freysa’s treasury should be allocated. The documentation describes thresholds, voice prioritization for active holders, and a path toward very broad participation. (framework.freysa.ai)
• Payments across the stack: The team plans to accept FAI for subscriptions to private AI apps and to use it as a medium for developer payments as more of the infrastructure and SDK get open‑sourced. (framework.freysa.ai)
• Compliance documentation: The framework links a MiCAR (EU crypto‑asset regulation) whitepaper for FAI, signaling intent to document the token’s characteristics for regulated markets. (framework.freysa.ai)

Ecosystem & Use Cases

Verifiable games that test human‑AI interaction

Freysa’s launch introduced a public challenge with clear rules: anyone could pay a message fee on Base to send a single message to Freysa. If the AI became convinced to transfer funds from a prize pool, the sender would win the pool. These “Acts” were designed as serious tests of social engineering, prompt design, and rule‑following, not just as entertainment. In one widely reported round, a user named p0pular.eth convinced the agent to release a prize pool of about 13.19 ETH, highlighting both the stakes and the value of red‑team testing for agent safety. (theblock.co)

The key idea behind these games is not the payout itself; it’s the public audit trail. Rules are hashed and attested inside a TEE, the game state is protected, and anyone can verify claims through the project’s public tools. That transparency lets the community inspect what went right—or wrong—and propose improvements. (freysa.ai)

Digital Twins and agent networks

Beyond games, the project’s long‑term vision is a network of “Digital Twins”—private personal AIs that represent their humans, remember context, and collaborate. Freysa acts as a coordinator for this network, while products like Enchanted and Silo give people a direct interface. Over time, the team plans “Pantheon,” a launchpad for creating many agents that operate in places people already spend time (messaging apps and social platforms), with verified tool use through MCP servers running in TEEs. (framework.freysa.ai)

Verifiable tool access

Because tools like email, cloud storage, and payments can be dangerous in the wrong hands, Freysa’s approach is to make the tool layer itself verifiable. Running MCP servers inside TEEs means an agent can, for example, send an email or check a document while producing an attestation others can check. This design opens a path for higher‑stakes use cases—identity checks, workflow approvals, or treasury actions—without asking users to trust a black box. (freysa.ai)

Advantages & Challenges

Freysa brings together ideas many crypto and AI builders care about:

• Strong verifiability. Key decisions and tool calls can be tied to TEE attestations, so outsiders can check what happened. (framework.freysa.ai)
• Clear, on‑chain coordination. The token and contracts live on Base, using familiar ERC‑20 standards and EVM tooling. (assets-cms.kraken.com)
• Real‑world testing. Public challenges create a steady stream of “red team” attempts that reveal where agents are brittle and where rules need hardening. (theblock.co)

At the same time, the project is intentionally experimental:

• Autonomy is hard. The same freedom that makes an agent useful also creates places to fail. The $47K win against Freysa in an early challenge shows why constant iteration is needed. (theblock.co)
• Complexity. Concepts like TEEs, attestations, and MCP can be new to many users and even many developers. The team’s docs help, but there is still a learning curve. (framework.freysa.ai)
• Governance design. Turning token‑holder input, chat signals, and technical attestations into fair decisions is a live research problem. The framework sketches options, but details will matter. (framework.freysa.ai)

Where to Buy & Wallets

Freysa AI can be purchased on major centralized exchanges and on Base‑native decentralized exchanges. As of recent listings, FAI is available on Coinbase Exchange, Gate.io, MEXC, Ourbit, KCEX, and XT.COM. On decentralized venues, FAI trades on Aerodrome (Base) and Uniswap v3 on Base. Always use the official contract: 0xb33Ff54b9F7242EF1593d2C9Bcd8f9df46c77935. (coingecko.com)

Because FAI is an ERC‑20 token on Base, it works with common EVM wallets. MetaMask, Coinbase Wallet, Rabby, and Trust Wallet can all connect to Base. If a wallet does not show Base by default, add it manually using chain ID 8453 (Base mainnet) per the official Base documentation. Hardware wallets like Ledger can connect through a compatible EVM wallet. (docs.base.org)

Regulatory & Compliance

Freysa operates as a tokenized software ecosystem on the Base network. The team links a MiCAR whitepaper for FAI, indicating an effort to describe the token’s characteristics under the European Union’s Markets in Crypto‑Assets regulation. While a whitepaper is not the same as regulatory approval, it shows attention to documentation that many jurisdictions expect. Availability on centralized exchanges also means those platforms apply their own listing standards and KYC/AML controls to users, which can differ by region. (framework.freysa.ai)

On religious compliance, there is no widely recognized, official shariah certification for Freysa AI at this time. Some observers view the token’s role—paying for access, coordinating governance, and rewarding participation in a transparent, rules‑based system—as generally aligned with utility use rather than interest‑based finance. Others note that the early public challenges included message fees and a prize pool, which could be interpreted differently across schools of thought depending on whether the activity is treated as a skill‑based contest or as gharar/maysir. In practice, many Muslim users look for clear certification from qualified scholars; until such a ruling exists, views are likely to vary. (theblock.co)

Future Outlook

Freysa’s roadmap points toward three developments.

First, more verifiable autonomy. Expect deeper use of Intel TDX, broader attestation coverage, and cleaner user tools for checking proofs. As the tool layer moves fully into TEEs, people and institutions may grow more comfortable letting agents touch sensitive systems. (freysa.ai)

Second, growth of agent networks. The planned Pantheon launchpad aims to make it easy to spin up many agents that live where users already are (Telegram, WhatsApp, X), each with verified toolkits. If successful, that could shift Freysa from a single‑agent experiment to a framework others can use to run their own, specialized agents. (framework.freysa.ai)

Third, token‑guided governance. The framework outlines how FAI could route human input—votes, chats, and on‑chain signals—into real decisions about policy stances, treasury use, and software changes. Doing this well will require careful mechanism design, but it is the core promise behind using a token in the first place. (framework.freysa.ai)

Summary

Freysa AI brings a fresh approach to AI‑crypto: pair an expressive agent with hard guarantees. TEEs and remote attestation make actions verifiable. MCP servers running inside enclaves make tools trustworthy. Base gives an EVM home where contracts, wallets, and exchanges already work. The FAI token ties it together with access, payments, and a path to shared governance.

The project’s public challenges showed both the fun and the difficulty of building autonomous systems. Being outwitted once is not the end; it is the point of an experiment that learns in the open. As Freysa moves from one agent to many, and from demos to durable apps, FAI’s role will be to invite people into that process—owning a piece of the network, shaping its rules, and helping prove that advanced AI can be both powerful and accountable. (theblock.co)