> ## Documentation Index > Fetch the complete documentation index at: https://docs.classiq.io/llms.txt > Use this file to discover all available pages before exploring further. # Quantum Engineering Interface > [!NOTE] > > **Beta:** The Quantum Engineering interface is currently in beta. ## Overview The Quantum Engineering interface is an agentic experience in the [Classiq platform](https://platform.classiq.io/). It carries a whole task from goal to result. Describe what you want to achieve, and the agent plans the work, writes a [Qmod model](/user-guide/modeling), synthesizes it into a quantum program, executes it, and analyzes the results. Each step produces an artifact you can open, inspect and continue from. To run the same workflow from your own agentic coding platform, such as Claude Code or Cursor, use the [Quantum Engineer plugin](/user-guide/ai/quantum-engineer-plugin). To get AI help while you write code in Classiq Studio, use the [AI agent in Studio](/user-guide/ai/studio-agent). ## Accessing the Interface > [!NOTE] > > The Quantum Engineering interface is available only to users with an active Classiq subscription. To open the Quantum Engineering interface, click **Try now** on the "Try our new agentic Quantum Engineering interface" banner at the top of the Classiq platform home page. ![The 'Try our new agentic Quantum Engineering interface' banner with its Try now button](https://mintcdn.com/classiq/1F2FC6_VjbsGxq0e/user-guide/ai/resources/quantum-engineering-interface-banner.webp?fit=max\&auto=format\&n=1F2FC6_VjbsGxq0e\&q=85\&s=eb5eece814ff01c12027c8d14d73a27c) The interface opens on a new session, with these elements: * **Prompt box:** Describe a circuit, algorithm or problem in the center of the screen. Click the **+** icon to upload a research paper you want to implement, a circuit image or diagram, or a code file such as a Qiskit circuit or a model. * **Threads:** Click **New Thread** in the left panel to start a separate conversation. * **More from Classiq:** Open [Classiq Studio](/user-guide/studio), the Library or the User Guide, or download the **AI Plugin**, which [installs the Quantum Engineer plugin](/user-guide/ai/quantum-engineer-plugin#installation) in your agentic coding platform. * **Legacy platform:** Click **Switch to the legacy platform experience** at the top right to go back to the non-agentic Classiq platform at any time. ![The Quantum Engineering interface main screen: a new session with a prompt box reading 'Enter your vision - describe a circuit, algorithm, or problem', a New Thread button, and links to Classiq Studio, Library, User Guide and AI Plugin](https://mintcdn.com/classiq/1F2FC6_VjbsGxq0e/user-guide/ai/resources/quantum-engineering-interface-main-screen.webp?fit=max\&auto=format\&n=1F2FC6_VjbsGxq0e\&q=85\&s=a504da7d864580b89dbf5c0f914700f0) ### Thread Context The agent keeps the context of the current thread so each step builds on the last: * **Attachments:** Papers, images and files you upload with **+**, such as a research paper, a circuit diagram, a Qiskit circuit or Qmod, stay available for the rest of the thread. * **Task state:** The synthesized program, execution results and analysis carry through the thread. You can say "now run it on a simulator" without restating the model. * **Starting a new thread:** Click **New Thread** to start fresh with no prior context. ## Running a Task 1. Type your goal in the prompt box, for example "Implement Grover's search", or click **+** to upload a paper, circuit image or code file to start from. 2. Click the arrow button to send. The agent replies with a plan, such as "model the circuit, synthesize it, then execute it on the simulator", and starts working through it. 3. As each step finishes, the agent explains what it did and adds an artifact card to the thread: **Code**, **Quantum Program**, **Execution Result** and **Analysis Report**. Expand **tool call** under a message to see exactly what the agent ran. 4. Click **Open** on any artifact card to view it in the panel on the right. Ask follow-up questions in the same thread to change the model, rerun a step, or explain a result. ### Artifacts Each artifact opens in the right panel with a button that takes it to the next step. You can click the button or ask for the same step in the chat. Use the dropdown at the top of the panel, such as **Code 1**, to switch between versions. * **Code:** The generated Qmod model in Python. Click **Synthesize** to build the quantum program. * **Quantum Program:** The synthesized circuit in the [interactive visualizer](/user-guide/analysis/visualization-of-quantum-programs). Click **Execute** to run it. * **Execution Result:** A histogram of measurement counts. Click **Analyse** to generate a report. * **Analysis Report:** A summary of the run with key figures and a per-state table of counts and probabilities. Click the download icon to save it. **Code** ![The agent explaining a generated Grover model next to the Code artifact, a Qmod Python file with a GroverVars QStruct, a @qperm predicate and a main function, with a Synthesize button](https://mintcdn.com/classiq/1F2FC6_VjbsGxq0e/user-guide/ai/resources/quantum-engineering-interface-code-artifact.webp?fit=max\&auto=format\&n=1F2FC6_VjbsGxq0e\&q=85\&s=8d2b2e629ae15387044581e7665cf184) **Quantum Program** ![The Quantum Program artifact showing the synthesized Grover circuit in the visualizer, with an Execute button](https://mintcdn.com/classiq/1F2FC6_VjbsGxq0e/user-guide/ai/resources/quantum-engineering-interface-quantum-program-artifact.webp?fit=max\&auto=format\&n=1F2FC6_VjbsGxq0e\&q=85\&s=7925f6e1f13963c71ca715f298f2dc0b) **Execution Result** ![The Execution Result artifact showing a histogram of measurement counts with two dominant states, 0110 and 1001, and an Analyse button](https://mintcdn.com/classiq/1F2FC6_VjbsGxq0e/user-guide/ai/resources/quantum-engineering-interface-execution-result-artifact.webp?fit=max\&auto=format\&n=1F2FC6_VjbsGxq0e\&q=85\&s=0b62924f0bd8ed6f56b96e1b04d321cb) **Analysis Report** ![The Analysis Report artifact titled Grover's Search End-to-End Results, with summary figures and a table of counts and probabilities per state](https://mintcdn.com/classiq/1F2FC6_VjbsGxq0e/user-guide/ai/resources/quantum-engineering-interface-analysis-report-artifact.webp?fit=max\&auto=format\&n=1F2FC6_VjbsGxq0e\&q=85\&s=0080a55eac89bde464020027013a002f) ## Writing Prompts The agent does best with a clear goal and room to choose the details, refined over follow-up messages. ### Examples These prompts show how far the agent takes a task from a single message: * "Implement Grover's search": models a Grover search, synthesizes it, runs it on the simulator, and produces an analysis report. * "Build a QAOA ansatz for MaxCut on a 6-node ring and run it on the simulator": generates the ansatz and runs the optimization. * "Implement the algorithm in this paper", with a paper uploaded: reads the paper and generates a model of the algorithm it describes. * "Implement quantum phase estimation for a 3-qubit unitary": generates the model and synthesizes the circuit. You can also ask follow-up and conceptual questions in the same thread: * "Why did you choose 2 Grover iterations?" * "Explain what this circuit does" * "Rerun it with 4,096 shots" ### Best Practices | Do | Don't | | - | - | | State the goal and how far to take it, such as "end to end" or "just synthesize" | Over-specify technical parameters in your first prompt. It over-constrains the agent and makes synthesis more likely to fail | | Start simple, then refine with follow-up messages in the same thread | Pack several unrelated tasks into one prompt. The agent handles one focused task more reliably | | Attach files or circuit images as starting points | Expect the agent to guess at code or data it can't see | | Open the artifacts and ask the agent to explain its choices | Assume a failed prompt means the use case is unsupported. Try a simpler version first | ## Capabilities With the Quantum Engineering interface, you can: * Go from a plain-language goal to code, circuit, results and a report in one thread * Generate Qmod models from descriptions, files or circuit images * Synthesize, execute and analyze directly from the artifact panel * Follow every step the agent takes, including its reasoning and tool calls * Refine any step with follow-up messages in the same thread * Download analysis reports to share results ## Limitations The interface is in beta, so treat the agent's output as a draft to review: * **Not every task will succeed.** Large or novel problems may produce incomplete or incorrect output. * **Results are not guaranteed to be correct.** Review the generated code and check results against known cases where you can. ## See Also * [AI with Classiq](/user-guide/ai): an overview of the AI options in Classiq. * [Quantum Engineer plugin](/user-guide/ai/quantum-engineer-plugin): the same workflow in Claude Code, Cursor, Codex or Copilot. * [Visualization of quantum programs](/user-guide/analysis/visualization-of-quantum-programs): how to read the circuit in the Quantum Program artifact. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.