CLARAI
The Yield Intelligence Layer for AI Agents.
Abstract
CLARAI is a yield intelligence layer for AI agents. It reads live DeFi yield data across Base and Ethereum, scores every stablecoin opportunity on risk-adjusted factors rather than raw APY, and serves that scored intelligence through a structured API that autonomous agents and the developers building them can consume directly.
CLARAI is deliberately non-custodial. It does not take deposits, route capital, hold funds, or execute transactions. It sits upstream of execution: it collects data, evaluates opportunities, scores risk and yield quality, explains its reasoning, and hands structured intelligence to whatever system is making the decision. The agent decides and acts; CLARAI informs.
"Raw data tells an agent what exists. CLARAI tells it what that data means."
The first intelligence product is YieldScore™ — a transparent 0–100 risk-adjusted score for stablecoin yield pools, already live. The longer-term direction is to become the broader DeFi intelligence layer autonomous agents consult before deploying capital on-chain.
The Problem
Finding yield is easy. Knowing which yield to trust isn't. An agent can find a pool paying 40% APY in seconds — but raw APY tells it almost nothing about whether that yield is real, safe, or about to vanish.
| What raw data shows | What it doesn't show |
|---|---|
| A high APY number | Whether the yield is sustainable or a temporary incentive about to dry up |
| A TVL figure | Whether liquidity is deep and stable, or quietly draining |
| A protocol name | Whether it's audited and established, or three weeks old |
| A current rate | Whether the APY has held steady or swings wildly week to week |
A 40% APY and a 4% APY can look identical to a machine reading raw data. One might be a genuine opportunity; the other might be gone tomorrow. Humans paper over this gap with intuition and experience. Agents can't. They need the judgment encoded as data.
Public data sources give an agent the raw numbers — APY, TVL, the mechanical facts. They don't score risk, weigh sustainability, or explain which opportunity is actually trustworthy. CLARAI is the interpretation layer on top: it turns fragmented data into scored, explainable intelligence an agent can act on.
The Opportunity
AI agents are evolving from assistants into economic actors — software that holds wallets, makes decisions, and moves capital on its own. As this shift accelerates, a growing share of on-chain capital will be allocated by machines rather than people.
The forecasts disagree on the size. Morgan Stanley and Bain measure near-term agentic transaction volume in the hundreds of billions by 2030; McKinsey's broadest estimate runs into the trillions. The number is contested — but every major analyst agrees on the direction. Autonomous agents will become a meaningful force in economic activity this decade.
Machines can't allocate capital the way humans do. They can't skim a forum thread or trust a gut feeling. They need financial intelligence built for machines: structured, scored, explainable, and available programmatically. That is the layer CLARAI is building.
This is an early and partly speculative market. CLARAI is positioned for where on-chain agent activity is heading, not for a market that exists at scale today. Forecasts cited are from the named institutions and are projections, not guarantees.
The Intelligence Engine
CLARAI's engine runs a simple, transparent pipeline:
| Stage | What happens |
|---|---|
| Ingest | Live stablecoin yield data is pulled across Base and Ethereum — APY, TVL, reward composition, and yield history for every tracked pool. |
| Score | Each pool is scored 0–100 on five risk-adjusted factors (see Methodology). High APY alone does not win; unsustainable yield is penalised. |
| Classify | Each pool is assigned a machine-readable classification band and a set of reason tags derived from its score breakdown. |
| Serve | Scored, classified intelligence is served through a structured API — filterable, explainable, and designed for machine consumption. |
Crucially, the output is interpreted intelligence, not raw data passed through. Every score carries its reasoning, so an agent can inspect why a pool scored what it did rather than trusting an opaque number.
YieldScore™ Methodology
YieldScore is a transparent 0–100 score. It is not a mysterious AI-generated number — it is the sum of five weighted factors, and every component is exposed in the API response so it can be inspected and verified.
| Factor | Max | What it measures |
|---|---|---|
| Audit status | 25 | Whether the protocol is audited and established |
| TVL depth | 25 | Total value locked — deeper, more stable liquidity scores higher |
| Protocol age | 20 | How long the protocol has operated without incident |
| APY sustainability | 20 | Very high APY scores lower — unsustainable or incentive-driven yield is penalised |
| APY stability | 10 | Deviation from the 30-day mean APY — stable yield scores higher |
Scores map to classification bands an agent can branch on: 80+ LOW_RISK_YIELD, 60–79 MODERATE_RISK_YIELD, 40–59 ELEVATED_RISK_YIELD, below 40 HIGH_RISK_YIELD. Each band maps directly to a score threshold, so the classification is always traceable back to the number behind it.
Reselling raw data is not defensible — anyone can do it. CLARAI's value is the scoring layer and, over time, a measurable track record: how pools CLARAI rated highly actually performed. That historical accuracy is what will make the intelligence credible, and it compounds from launch onward.
Built for AI Agents
CLARAI's intelligence is served as a structured API, with a native MCP (Model Context Protocol) server on the roadmap so agents can call CLARAI as a tool directly. Responses are machine-native: consistent typed fields, classification enums, and reason tags — not human prose reformatted as JSON.
| Capability | Description |
|---|---|
| Scored pools | Every tracked pool with its YieldScore, classification, and full breakdown |
| find_opportunities | Constraint-based query — asset, chain, min TVL, min score, risk band |
| explain_score | Structured explanation of a single pool: each factor, its contribution, and the reasons |
| MCP server | Roadmap — native tool access for MCP-compatible agents and assistants |
"Built for AI agents" is intended to be testable, not a slogan. The measure of success is a developer connecting an agent to CLARAI and having it reason over real, scored intelligence — not raw data it has to interpret itself.
CLARAI never holds, moves, or routes funds. There is no vault, no deposit, no execution. This removes smart-contract risk to user capital entirely — there is nothing to exploit and nothing to trust. The consuming agent takes every action itself.
$CLARAI Token
CLARAI is built product-first. The token exists — deployed and verified on Base — but it is deliberately not the pitch. The intelligence is. The token's role is being designed around how agents and developers actually consume CLARAI, rather than invented ahead of usage.
| Property | Detail |
|---|---|
| Supply | 1,000,000,000 fixed — no minting, no burning |
| Transfer tax | None |
| Ownership | Renounced — contract is immutable |
| Chain | Base (ERC-20) |
| Market status | No liquidity pool — not tradeable. Any "tradeable $CLARAI" today is a scam. |
Intended utility. As machine-to-machine payments mature, agents and developers will pay for CLARAI's intelligence per query, in stablecoins — keeping access frictionless. Holding $CLARAI is intended to unlock preferential access and pricing for heavy users: the more a consumer relies on CLARAI, the more rational it becomes to hold the token. These mechanics will be finalised around real usage before any launch.
No revenue-sharing, no buyback mechanism, no claim that token value tracks business performance. Those designs carry regulatory weight and are not promised here. $CLARAI is intended as an access and utility asset, not a claim on revenue.
Token Allocation
Total supply of 1,000,000,000 $CLARAI, distributed as deployed on-chain:
| Allocation | Amount | Share | Status |
|---|---|---|---|
| Community / Ecosystem | 600,000,000 | 60% | For distribution, incentives, ecosystem growth |
| Founder | 150,000,000 | 15% | Locked on Unicrypt, vesting — lock to be extended before launch |
| Liquidity | 100,000,000 | 10% | Reserved for the eventual liquidity pool |
| Treasury | 100,000,000 | 10% | Governance-controlled |
| Advisors / Contributors | 50,000,000 | 5% | Reserved for future contributors, with vesting |
The founder allocation is locked on-chain and publicly verifiable via Unicrypt. As a long-term commitment, the lock terms are intended to be extended before any token launch — the team should be rewarded as the product is built, not before.
Chain — Base
CLARAI is built on Base, Coinbase's Ethereum L2. Base offers low fees, deep stablecoin liquidity, native USDC, and a growing ecosystem of the exact protocols CLARAI scores. It is also where much of the early on-chain agent and x402-style machine-payment activity is emerging — making it the natural home for agent-facing financial infrastructure.
Roadmap
| Phase | Focus |
|---|---|
| Live now | YieldScore live across Base + Ethereum. Agent API with scoring, find_opportunities, and explain_score. |
| Next | Native MCP server. Historical score logging to build a measurable track record. |
| Later | Additional intelligence primitives — protocol risk, stablecoin quality, strategy scoring. Pay-per-query access and token utility finalised around real usage. |
| Vision | The broader DeFi intelligence layer autonomous agents consult before deploying capital on-chain. |
Roadmap items beyond "live now" are intentions, not commitments or dates. They will be built in order of what makes CLARAI genuinely more useful to agents — not to pad a timeline.
Risks
These risks are disclosed honestly, so participation — if any — is informed.
| Risk | Description |
|---|---|
| Early market | Autonomous agents consuming yield intelligence at scale is a nascent, partly speculative market. It may develop slower than expected, or differently. |
| Data dependency | CLARAI's intelligence is derived from public data sources. Errors, gaps, or outages upstream affect the scores. Scores may be incomplete or delayed. |
| Scoring limitations | No scoring model is perfect. A high YieldScore is not a guarantee of safety, and CLARAI may not anticipate novel risks or exploits. |
| Token not live | $CLARAI has no liquidity pool and is not tradeable. There is no guarantee it ever launches, or of any token value if it does. |
| Execution risk | Roadmap features including the API expansion and MCP server are intentions, not commitments, and may change or not ship. |
| Regulatory | The regulatory treatment of tokens and agent infrastructure is evolving and could affect the project. |
CLARAI provides research and information, never financial advice. Nothing here is a recommendation to buy, hold, or allocate anything. Conduct your own research.
This litepaper is provided for informational purposes only and does not constitute financial, investment, or legal advice. It is not an offer or solicitation to buy or sell any financial instrument. CLARAI provides risk-adjusted yield intelligence derived from public data, for research purposes only — it is not a recommendation to deposit, buy, or allocate funds, and CLARAI never takes custody of or routes user funds. Scores are informational and may be incomplete or outdated. The $CLARAI token has no liquidity market and is not tradeable; any claim otherwise is fraudulent. Information herein is subject to change without notice, and roadmap items are intentions rather than commitments. The team makes no guarantees regarding token value, product performance, or future development. Crypto assets are highly speculative and volatile; you may lose the entirety of any funds deployed. This document has not been reviewed or approved by any regulatory authority. Residents of jurisdictions where participation is restricted should not interact with CLARAI.