How the Agent works

The AI Agent follows a structured execution loop:

  1. Intent Understanding The agent interprets user intent expressed via text or voice.

  2. Action Planning The intent is translated into a sequence of onchain actions (e.g. swaps, bridges, deployments).

  3. User Authorization The user reviews and signs the proposed execution.

  4. Onchain Execution Actions are executed across chains and protocols.

  5. Monitoring & Feedback Execution results are tracked and surfaced to the user.

At every step, the user remains in control.


Interaction Modes

Users can interact with the NOYA AI Agent through:

  • Text Precise, structured, or exploratory commands.

  • Voice Natural, conversational interaction for faster execution and accessibility.

Both modes result in the same underlying execution flow and security guarantees.

The text and voice agent maintains contextual memory and is available across all NOYA products, allowing users to carry insights, preferences, and workflows seamlessly between intelligence, execution, vaults, and prediction markets.


Supported Actions

The NOYA AI Agent supports a growing set of onchain actions, including:

  • Asset swapping

  • Cross-chain bridging

  • Portfolio and position management

  • Capital deployment into Omnivaults

In addition to execution, the AI Agent provides institutional-grade intelligence and research capabilities at no cost.

Users can query the agent about:

  • DeFi yield opportunities, strategy design, and risk trade-offs

  • Token fundamentals, protocol architecture, revenue models, and governance

  • Prediction markets, probabilities, and scenario analysis

  • Onchain activity, capital flows, narratives, and systemic risk factors

NOYA’s intelligence layer is designed to be detailed, extensive, and exhaustive, mirroring the depth of professional research desks and crypto funds. Analysis spans fundamentals, market structure, incentives, risks, and recent developments, synthesized into clear, actionable insights.

This allows users to carry an advanced crypto fundamental researcher in their pocket, accessible through natural language or voice, without relying on fragmented dashboards or external tools.

The agent abstracts the complexity of both research and execution, while maintaining transparency.


Safety & Confirmations

The AI Agent operates with human-in-the-loop safeguards.

Before execution:

  • All actions are presented clearly to the user

  • Required approvals and signatures are explicit

  • Execution parameters are visible and reviewable

The agent cannot move funds autonomously. Execution only occurs after explicit user authorization.


Limitations

The AI Agent:

  • Does not provide financial advice

  • Does not act without user consent

  • Operates within predefined permissions and constraints

Users are responsible for reviewing actions before execution.

Last updated