Brian - Documentation
  • 🖐️Welcome
  • 👀Prompt-to-earn
  • 🤖Why an Intent Recognition Engine?
  • 🌐Brian AI models
  • 📜Prompt Guide
  • 🧠Use Cases
  • 💻Powered by Brian
  • ⁉️FAQ
  • Brian API
    • ⚙️APIs
      • 💻API Swagger
      • 💸/transaction
        • ⛓️Actions, Networks, and Tokens supported
          • Deposit and Withdraw to/from DeFi protocols
        • Transactions flow
          • EVM
          • Solana
          • Starknet
      • 📚/knowledge
        • 📑Knowledge Boxes
        • 🐝Public Embeddings
      • 🧙‍♂️/agent
      • 🧑‍🔬/parameters-extraction
      • 👨‍💻/smart-contract (alpha)
      • 📰/networks
      • ℹ️/actions
    • 🥳What's New?
    • 🔑API key
  • Brian SDK
    • 📚Brian Typescript SDK
  • AI AGENTS
    • 🤖LangChain<>Brian SDK
  • Brian App
    • 🍀Overview
    • 📤Send Transactions
    • 🙏Ask Brian
    • 🚚Deploy Smart Contracts
  • Misc.
    • Telegram Dev community
    • Website
    • Github
    • Medium
    • Twitter
    • HuggingFace
Powered by GitBook
On this page
  • ‣ AI Agent
  • ‣ Multiple AI Agents collaboration
  • ‣ Intent based transaction app
  • ‣ AI Powered Frame
  • ‣ XMTP AI Powered Bot
  • ‣ XMTP Multi AI Powered Bot in a Group chat
  • ‣ Telegram AI Powered Bot
  • ‣ Telegram mini app
  • ‣ Wearable/hardware devices
  • ‣ Extend Brian API capabilities (using the parameters-extraction endpoint)
  • ‣ Voice recognition
  • ‣ Stable coin payments
  • ‣ Hackathon assistant
  • ‣ Metamask Snap
  • ‣ Browser Extensions
  • ‣ Token bound Account

Use Cases

Some Use Cases devs can build on top of the Brian API and SDKs

PreviousPrompt GuideNextPowered by Brian

Last updated 4 months ago

‣ AI Agent

AI Agent is the hottest topic in the web3 right now. Build an AI Agent that can execute transactions independently or recommend transactions to the user.

Examples could be:

  • DCA: executes transaction recurrently.

  • Limit Order: executes transaction when a specified price is reached.

  • DeFi Optimizer: dynamically invest on DeFi protocols according to some risk/return preferences.

  • Portfolio Manager: balance asset allocation, yield generation, and risk management making data-driven investments.

Other more hardcore ideas include Agents that manage twitter/farcaster accounts by posting and executing transactions independently. Examples include:

Suggested Brian endpoints:

  • /transaction

  • /knowledge

  • /agent

  • /parameters-extraction


‣ Multiple AI Agents collaboration

Push AI Agents to the next step and build an app/framework/etc.. where multiple AI Agents collaborate each other executing transactions independently or recommending transactions to the user. You can find some ideas at the previous point (AI Agent).

For example, you could build a collaborative multi Agents system using LangChain or OpenAI Swarm library that leverages the Brian API.

Take inspiration from:

Suggested Brian endpoints:

  • /transaction

  • /knowledge

  • /agent

  • /parameters-extraction

  • In-app AI Assistant

    Embed an AI-powered assistant within an app/wallet/protocol, enabling users to execute transactions, gather market insights, ask information about the app or protocol, or receive personalized investment recommendations through conversational prompts. This assistant could even leverages real-time data and context to provide streamlined user guidance.


    Suggested Brian endpoints:

    • /transaction

    • /knowledge

    • /agent

    • /parameters-extraction


‣ Intent based transaction app

A transaction app designed to understand and respond to user intent directly from natural language prompts, interpreting commands like "invest in low-risk DeFi" or "set up daily purchases" to automate the desired actions with minimal input.

Suggested Brian endpoints:

  • /transaction

  • /knowledge

  • /agent

  • /parameters-extraction


‣ AI Powered Frame

Suggested Brian endpoints:

  • /transaction

  • /knowledge

  • /agent

  • /parameters-extraction


‣ XMTP AI Powered Bot

Suggested Brian endpoints:

  • /transaction

  • /knowledge

  • /agent

  • /parameters-extraction


‣ XMTP Multi AI Powered Bot in a Group chat

Imagine having your personal agents in a group and using them on demand to perform different types of tasks

Suggested Brian endpoints:

  • /transaction

  • /knowledge

  • /agent

  • /parameters-extraction


‣ Telegram AI Powered Bot

Build AI powered Telegram bots that integrates natural language prompts in a chat for making transactions or requesting information.

Suggested Brian endpoints:

  • /transaction

  • /knowledge

  • /agent

  • /parameters-extraction


‣ Telegram mini app

Build AI powered Telegram mini app that integrates natural language prompts for making transactions or requesting information.

Suggested Brian endpoints:

  • /transaction

  • /knowledge

  • /agent

  • /parameters-extraction

‣ Wearable/hardware devices

Use Brian API within wearable applications (e.g., apple watch). Provides timely, voice-activated financial insights, make transactions and give market recommendations directly to a smartwatch or similar device, keeping users connected and responsive to market changes on the go.

Suggested Brian endpoints:

  • /transaction

  • /knowledge

  • /agent

  • /parameters-extraction


‣ Extend Brian API capabilities (using the parameters-extraction endpoint)

Extend the current /transaction and /knowledge capabilities using the /parameters-extraction endpoint to build new custom logic. You could implement a new solver, a new chain, a new DeFi protocol and so on. Just call the /parameters-extraction, take the parameters extracted and build your own logic with it.

Suggested Brian endpoints:

  • /parameters-extraction


‣ Voice recognition

Currently, the Brian API uses text inputs. Build AI-enabled voice recognition allowing users to execute transactions or ask information simply by speaking commands, providing a hands-free and accessible way to interact with their assets. Attention: the solution must be robust! it's not enough to use a model to turn voice into text. Why? Standard voice recognition models don't know web3 vocabulary and would work poorly.

Suggested Brian endpoints:

  • /transaction

  • /knowledge

  • /agent

  • /parameters-extraction


‣ Stable coin payments

Build AI agents/apps that facilitate and optimize stablecoin payments, whether for DeFi investments, cross-border transfers, or routine transactions proposing. Leverage natural language for proposing new ways of interacting with stablecoins.

Suggested Brian endpoints:

  • /transaction

  • /knowledge

  • /agent

  • /parameters-extraction


‣ Hackathon assistant

Build an app/service specialized for hackathons, guiding participants through the different rules, prizes and so on. If you need it, ask the Brian team to create a new custom knowledge box that you can leverage using /knowledge endpoint.

You might think about integrating this service within sites but more interestingly within Notion pages.

Suggested Brian endpoints:

  • /knowledge

  • /agent


‣ Metamask Snap

Build a Metamask Snap integrating natural language prompts extending Metamask standard wallet functionalities for making transactions or requesting information.

Suggested Brian endpoints:

  • /transaction

  • /knowledge

  • /agent

  • /parameters-extraction


‣ Browser Extensions

Build a browser extension providing in-situ AI assistance for transaction management, data analysis, and decision-making, transforming the browser into an interactive, AI-driven Web3 toolkit. You could have this extension that leverages a browser wallet (eg. Metamask) for making web3 transactions or perhaps the extension could have a built-in wallet? 🤷

Suggested Brian endpoints:

  • /transaction

  • /knowledge

  • /agent

  • /parameters-extraction


‣ Token bound Account

Suggested Brian endpoints:

  • /agent

  • /parameters-extraction

Build frames (using the ) that integrates natural language prompts. An example is:

Build AI powered XMTP bots that integrates natural language prompts in a chat for making transactions or requesting information. An example is brian-agent.eth (try on , the main XMTP client).

Build an XMTP group chat with multiple AI powered Bots/Agents that integrates natural language prompts in a chat for making transactions or requesting information. An example is .

You can check the SnappyAI example here:

Build an app leveraging the . Imagine an app that lets you create a smart account controlled by an NFT. Chat with your NFT in natural language to perform transactions, request information etc.. Propose new NFT user experiences and increase their usefulness. Maybe the NFT can have its own personality or customizable features?

🧠
tee_hee_he
terminal of truths
Luna
elefant
aethernet
lincoln.base.eth on Twitter / X
open frame standard
https://github.com/brian-knows/brian-open-frame
Converse
https://ethglobal.com/showcase/power-agents-djqmb[b](https://ethglobal.com/showcase/power-agents-djqmb)
https://ethglobal.com/showcase/snappyai-n2igg
ERC-C6551 Token Bound Account Standard