> For the complete documentation index, see [llms.txt](https://docs.ardentresearch.xyz/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.ardentresearch.xyz/readme.md). # About Ardent AI Research Welcome to the official Testnet documentation from Ardent AI Research. This guide is written for teams integrating with the hosted Ardent API at `https://api.ardentresearch.xyz`. ## Who we are Ardent AI Research is an R\&D lab building infrastructure for agent autonomy. Our vision is to make it easy for AI agents to securely execute real tasks such as payments, on-chain actions, and service calls without centralized gatekeepers. Our first product, Jusso, is a new generation of autonomous infrastructure for onchain action. The Jusso Beta is coming soon on Base; the platform documented here is the deployed Testnet preview. ## What is an AI agent blockchain execution platform An AI agent blockchain execution platform is the trust and execution layer between an agent's intent and real-world settlement. In practice, it provides: 1. Wallet abstraction and account lifecycle management. 2. Deterministic execution and transaction routing across chains. 3. Sponsored testnet execution through ERC-4337 paymasters. 4. Policy, observability, and status tracking for async execution. Without this layer, teams typically stitch together fragile components and inherit security and reliability risk at the exact point where agents touch value. ## The problem we solve Many agent systems can reason, plan, and generate actions, but they still fail at the final mile: safe and reliable execution. Common blockers include: 1. Wallet lifecycle complexity and signing constraints. 2. Non-deterministic transaction handling across chains and RPC environments. 3. Safe gas sponsorship for API-triggered execution. 4. Poor observability across asynchronous execution flows and retries. 5. Security boundaries between agent logic, keys, relayers, and external services. ## Our goal Our goal is to make autonomous execution safe by default, programmable, and easy to integrate. Teams should focus on product logic and agent behavior while Ardent handles execution correctness, gas sponsorship, and lifecycle reliability. ## AI agent blockchain execution platform The AI agent blockchain execution platform is our hosted execution surface for production agent workflows. It enables your software agent to execute EVM transactions through an ERC-4337 smart wallet without running your own relayer or custom broadcasting stack. At a high level, the platform provides: 1. Deterministic smart wallet resolution per agent. 2. Pre-flight simulation before broadcast. 3. Paymaster-sponsored UserOperation gas on supported testnets. 4. Standardized request lifecycle tracking via `request_id` and `GET /status/:id`. 5. Hosted API ergonomics with webhook-friendly asynchronous completion. For hosted users, the integration flow is straightforward: 1. Generate a public self-service API key. 2. Resolve or provision an agent wallet address. 3. Fund the agent wallet as needed. 4. Simulate or execute transactions. 5. Track execution state until completion. ## Who this guide is for This documentation is for: 1. AI agent developers and engineers integrating on-chain execution. 2. Product engineers building agent workflows. 3. DevOps engineers operating production integrations. ## Core concepts Before you begin, keep these platform concepts in mind: 1. API keys are customer scoped and required for protected endpoints. 2. Every agent is mapped to a deterministic smart wallet. 3. Every execution is simulated before it can enter the queue. 4. `GET /status/:id` is the source of truth for lifecycle state. ## Base URL Use the hosted API base URL: ```bash https://api.ardentresearch.xyz ``` ## Agent integration Ardent AI Research ships an integration pack that makes it fast to connect any developer, script, or AI tool to the Testnet. It includes: 1. A zero-dependency CLI (`ardent`) with a one-line installer. 2. An MCP server that exposes all platform tools to Codex, Claude Desktop, ChatGPT Desktop, Cursor, Windsurf, and Hermes Agent. 3. An OpenAPI 3.1 spec for ChatGPT custom actions and code generators. The installer auto-patches desktop AI configs and preserves any existing API key. See [Agent Integration](/agent-integration.md). ## Next step Start with [Getting Started](/getting-started.md). For Sepolia token details and testnet funding guidance, see [Testnet Guide](/testnet-guide.md). --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.ardentresearch.xyz/readme.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.