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

# Neum Client

> Interact with the Neum Cloud

The `NeumClient` allows you to interact with the Neum AI Cloud through a python SDK. It supports the APIs listed below as methods in the client.

## Initialize client

<Note>You will need a Neum AI Cloud API Key to initialize the client. Go to [dashboard.neum.ai](dashboard.neum.ai) to create an account and get an API Key</Note>

```python Initialize NeumClient
from neumai.Client.NeumClient import NeumClient
neumClient = NeumClient(api_key = <INSERT NEUM AI KEY>)
```

## Create pipeline

Configure your pipeline first using the [Neum AI connectors](/connectors/introduction).

```python Create pipeline
from neumai.Pipelines import Pipeline
pipeline = Pipeline(source=[...], embed=... , sink=...)
pipeline_id = neumClient.create_pipeline(pipeline=pipeline)
```

## Get pipeline

Retrieve the pipeline configuration and latest run information:

```python Get pipelines
print(neumClient.get_pipeline(pipeline_id=pipeline_id))
```

## Search pipeline

Query the data indexed using the pipeline configuration:

```python Create pipeline
print(neumClient.search_pipeline(pipeline_id=pipeline_id, query="What is ....", num_of_results=3, track=False, requested_by="John Doe"))
```

We also support search against a specific file within a pipeline using the `search_file` method. Ex. if you ingested an S3 bucket with 10 files, you can pick one and only search against it.

## Provide feedback on a retrieval

Capture feedback for a given search result. (Must have enabled retrieval tracking.)

```python Capture feedback
provide_retrieval_feedback(pipeline_id=pipeline_id, retrieval_id=retrieval_id, status="Add feedback ex. Good or Bad")
```

## Get search retrievals for a pipeline

Query the retrieval events against a pipeline (Must have enabled retrieval tracking.)

```python Get retrievals for pipeline
print(neumClient.get_retrievals_by_pipeline_id(pipeline_id=pipeline_id))
```

## Additional methods provided

* `trigger_pipeline`: Trigger a pipeline to run
* `get_pipelines`: Get a list of all pipelines created by a user
* `get_pipeline_runs`: Get a list of all the runs for a given pipeline
* `get_pipeline_run`: Get a specific run for a pipeline
* `get_files`: Get that statuses for all files available within a pipeline
* `get_file`: Get the status for a specific file within a pipeline
* `get_retrievals_by_file_id`: Get retrieval events for a specific file within a pipeline. (Must have tracking enabled)
* `get_retrievals_by_pipeline_id_user_id`: Get retrieval events for a user within a pipeline. (Must have tracking enabled and passed a `requested_by` parameter)
* `get_retrievals_by_file_id_user_id`: Get retrieval events for a user within a file. (Must have tracking enabled and passed a `requested_by` parameter)
* `get_retrievals_by_user_id`: Get retrieval events for a user across pipelines. (Must have tracking enabled and passed a `requested_by` parameter)