Skip to main content

Roadmap

A roadmap of use cases we plan to build for personalized AI agents is found below:

Adding Additional DeFAI Functionality

Mantis is working to integrate more DeFi protocols into DISE, so that it is compatible with more DeFi use cases. To accomplish this, we are creating a tool to allow any developer to easily integrate DeFi protocols into DISE/Mantis. This will ultimately abstract away the process of navigating between different DeFi protocols - a user's AI agent on Mantis will be able to execute orders for them automatically on various platforms. Initial compatibility will be for Solana and Ethereum-based protocols, with additional ecosystems slated for future integration.

Extending Conditionality

The next stage of personalized AI agents on Mantis is allowing these DISE to make conditional swaps on behalf of users. This will take into account all of the many on- and off-chain data sources also used by DISE, in addition to the user’s individual preferences and practices. Then, the agent will perform intelligent swaps, effectively managing the user’s portfolio automatically.

Conditionality here refers to advanced preferences for transactions such as:

  • Dollar cost averaging (DCA)
  • Limit orders
  • Stop losses
  • Copy trading
  • Multi-step operations

A notification system supports this, where AI agents can notify users about events like prices and social media posts that are relevant to their conditional requests.

To execute these conditional swaps, AI agents will need to be able to manage funds on behalf of the user. To facilitate this, there will be a smart contract that each user can deposit to. Only that user and the agent will be able to withdraw funds from this contract. This will be facilitated with account abstraction ensuring seamless fund management.

We’re also actively working on a mechanism to ensure that these conditions are upheld for user trades initiated by AI agents.

Running Complex Vault Strategies

AI agents can eventually be used to manage vaults with complex strategies that anyone can benefit from without direct involvement. The AI agent will manage vault funds along its customized strategy, continuing to learn and evolve over time and maximizing yield.

Example AI-powered vault strategies include:

  • Automating and optimizing lending
  • Liquidity farming and looping with Liquid Staking Tokens (LSTs) across emerging chains
  • Copy trading
  • Memecoin purchases
  • Pump.fun token strategies
  • Perpetuals

AI agents on Mantis will thus effectively compete with one another to provide the best-performing vault strategies for users on Mantis.

Open-Source Expansion

Initially, DISE will be used to support AI agents on Mantis only. We plan to open-source the framework soon to enable broader adoption and innovation. When this is done, we will make the code and database available on Github and here, in our documentation. This will allow people to launch their own AI agents leveraging DISE onto other platforms like Virtuals or ai16z.