> ## Documentation Index
> Fetch the complete documentation index at: https://docs.zeroruntime.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Record, train and run a policy

> Go from recorded demos to a trained policy that drives your arm.

There are four steps. You run scripts for the first and last. The middle two
happen in the [dashboard](https://app.zeroruntime.ai), under **Robotics**. No
code needed there.

| Step | Where | How | Guide |
| - | - | - | - |
| 1. Record | robot and leader machines | [`recording/`](https://github.com/ZeroRuntimeAI/zrt-teleops-examples/tree/main/recording) scripts | [Cloud recording](https://docs.zeroruntime.ai/robotics/guides/cloud-recording.md) |
| 2. Make a dataset | dashboard | **Recordings → Convert to dataset** | [Datasets](https://docs.zeroruntime.ai/robotics/guides/datasets.md) |
| 3. Train | dashboard | **Datasets → Train** | [Fine-tuning](https://docs.zeroruntime.ai/robotics/guides/fine-tuning.md) |
| 4. Run | robot, and a machine with a GPU | [`inference/`](https://github.com/ZeroRuntimeAI/zrt-teleops-examples/tree/main/inference) scripts | [Running a policy](https://docs.zeroruntime.ai/robotics/guides/running-a-policy.md) |

Two things carry through every step:

* **The task text**, like `pick up the cube`. You set it when you record. The
  policy learns to do that task.
* **The Training ID.** You copy it from the dashboard after training. The
  inference script uses it to download your policy.

## Before you start

* Two SO-101 arms, calibrated. See [SO-101](https://docs.zeroruntime.ai/robotics/integrations/lerobot/so101.md).
* An auth token and a room id. Use the same ones on both machines. Cloud
  recording is billed to the token's account.
* Python 3.12 or newer, and the examples repo:

```bash theme={null}
git clone https://github.com/ZeroRuntimeAI/zrt-teleops-examples.git
cd zrt-teleops-examples
python3 -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
pip install "zeroruntime-teleops[lerobot]>=0.1.2"
```

Then go to step 1: [Cloud recording](https://docs.zeroruntime.ai/robotics/guides/cloud-recording.md).

## Related

> [card] [Cloud recording](https://docs.zeroruntime.ai/robotics/guides/cloud-recording.md)
>
> Step 1: record demos.

> [card] [Datasets](https://docs.zeroruntime.ai/robotics/guides/datasets.md)
>
> Step 2: turn a recording into a dataset.

> [card] [Fine-tuning](https://docs.zeroruntime.ai/robotics/guides/fine-tuning.md)
>
> Step 3: train a policy.

> [card] [Running a policy](https://docs.zeroruntime.ai/robotics/guides/running-a-policy.md)
>
> Step 4: let the policy drive the arm.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.