Ask AI

Ask AI is Avana's conversational interface for questions about markets and positions. A request identifies the market, asset, or position to examine, and follow-up questions can clarify the inputs and assumptions. The response explains the available information; protocol checks still determine whether a requested transaction can proceed. Wallet-specific questions depend on the selected wallet and network. Public market data describes the market, while balances and positions belong to an account on a particular network. A missing balance, price, or route leaves the associated calculation incomplete.

Ask AI uses Aave MCP alongside Avana's own retrieval system and product implementation. The Model Context Protocol provides a structured interface to external tools. Aave's server exposes v3 and v4 market information, including reserves, rates, available liquidity, and protocol structure. This gives Ask AI Aave-specific context for questions that also involve Avana markets or positions. See Aave for Agents and the Aave MCP documentation for the upstream service and its interfaces.

Aave data retains its network, protocol version, and reserve context. It does not establish that the same collateral or debt asset is enabled in an Avana Borrow Spoke. Avana still applies its own pool admission, LP valuation, and account checks. A capability offered by the upstream MCP server is also not automatically an action exposed by Ask AI: available actions depend on Avana's integration and the supported product flows.

Avana's retrieval-augmented generation (RAG) searches its protocol knowledge for material relevant to a question and supplies that material as context for the answer. Retrieved passages explain the architecture, collateral rules, liquidation process, and product workflows, with source information that lets the explanation be traced to its documentation. This supplies protocol knowledge rather than live prices; rates and balances come from the relevant data sources.

Avana's own position data and calculation logic provide the product-specific context. A borrowing question can combine Aave market information, retrieved explanations of LP collateral, and calculations for the selected position's debt and borrowing capacity. Each input has a different role: a documentation passage explains a rule, a market response reports observed conditions, and a position calculation applies the relevant inputs to an account or scenario.

Market comparisons can include borrowing costs, available liquidity, and collateral requirements. A displayed yield is only one input: debt interest and changes in the LP position also affect the result. Input sources, timestamps, and scenario assumptions provide the context needed to interpret the comparison.

Analysis and transaction execution are separate steps. For a supported action, the relevant product flow presents the account, network, assets, amounts, and costs for review and requests any required wallet approval. Submission is followed by confirmation and an update to the resulting position.