diff --git a/doc/source/index.rst b/doc/source/index.rst
index f1ac436..f4e23c5 100644
--- a/doc/source/index.rst
+++ b/doc/source/index.rst
@@ -14,7 +14,7 @@ Contents:
 
    readme
    installation
-   usage
+   usage/index
    v1-api-documentation
    authentication-documentation
    contributing
diff --git a/doc/source/installation.rst b/doc/source/installation.rst
index 5d2d299..887ebac 100644
--- a/doc/source/installation.rst
+++ b/doc/source/installation.rst
@@ -2,6 +2,15 @@
  Installation
 ==============
 
+.. note::
+
+    There has not been a stable release of python-cratonclient to PyPI yet. To
+    install the current version in development use:
+
+    .. code::
+
+        pip install git+https://git.openstack.org/openstack/python-cratonclient
+
 At the command line::
 
     $ pip install python-cratonclient
diff --git a/doc/source/usage.rst b/doc/source/usage/authentication.rst
similarity index 53%
rename from doc/source/usage.rst
rename to doc/source/usage/authentication.rst
index 2708cfa..dbbfeaa 100644
--- a/doc/source/usage.rst
+++ b/doc/source/usage/authentication.rst
@@ -1,24 +1,8 @@
-=======================
- Python API User Guide
-=======================
+.. _usage-auth:
 
-Once you have installed ``python-cratonclient``, there are a few things you
-need to get started using the Python API:
-
-#. You need to know how to authenticate to the Craton API you wish to talk to
-
-   Some Craton API services will be deployed using Craton's in-built
-   authentication system while others may use Keystone.
-
-#. You need your credentials
-
-#. You need the location of your Craton API service
-
-Let's cover authentication first:
-
-
-Authenticating to Craton
-========================
+==========================
+ Authenticating to Craton
+==========================
 
 There are two ways to authenticate to Craton:
 
@@ -71,9 +55,6 @@ Then, we need to do the following:
 
 .. code-block:: python
 
-    from keystoneauth1.identity.v3 import password as password_auth
-    from keystoneauth1 import session as ksa_session
-
     from cratonclient import auth
     from cratonclient.v1 import client
 
@@ -89,46 +70,3 @@ Then, we need to do the following:
         session=craton_session,
         url=URL,
     )
-
-
-Communicating with Craton
-=========================
-
-Now that you've configured your authentication method, you can interact with
-your ``craton`` object like so:
-
-.. code-block:: python
-
-    for region in craton.regions.list():
-        print('Region {} contains:'.format(region.name))
-        for host in craton.hosts.list(region_id=region.id):
-            print('    {}'.format(host.name))
-
-
-The Craton API has the following resources:
-
-- Cells
-
-- Hosts
-
-- Network Devices
-
-- Network Interfaces
-
-- Networks
-
-- Projects
-
-- Regions
-
-- Users
-
-Of these:
-
-- Cells
-
-- Hosts
-
-- Regions
-
-Are implemented.
diff --git a/doc/source/usage/cells.rst b/doc/source/usage/cells.rst
new file mode 100644
index 0000000..d602640
--- /dev/null
+++ b/doc/source/usage/cells.rst
@@ -0,0 +1,136 @@
+=====================
+ Using the Cells API
+=====================
+
+Here we will assume that we already have a
+:class:`~cratonclient.client.Client` instance configured with the appropriate
+authentication method (as demonstrated in :ref:`usage-auth`).
+
+Listing Cells
+-------------
+
+The Cells API implements pagination. This means that by default, it does not
+return all cells known to Craton. To ignore page limits and offsets, we can
+allow cratonclient to do handle pagination for us:
+
+.. code-block:: python
+
+   for cell in craton.cells.list():
+       print_cell_info(cell)
+
+By default :meth:`~cratonclient.v1.cells.CellsManager.list` will handle
+pagination for you. If, instead, you want to handle it yourself you will want
+to do something akin to:
+
+.. code-block:: python
+
+    first_page_of_cells = list(craton.cells.list(autopaginate=False))
+    marker_id = first_page_of_cells[-1].id
+    second_page_of_cells = list(craton.cells.list(
+        autopaginate=False,
+        marker=marker_id,
+    ))
+    marker_id = second_page_of_cells[-1].id
+    third_page_of_cells = list(craton.cells.list(
+        autopaginate=False,
+        marker=marker_id,
+    ))
+    # etc.
+
+A more realistic example, however, might look like this:
+
+.. code-block:: python
+
+    cells_list = None
+    marker = None
+    while cells_list and cells_list is not None:
+        cells_list = list(craton.cells.list(
+            marker=marker,
+            autopaginate=False,
+        ))
+        # do something with cells_list
+        if cells_list:
+            marker = cells_list[-1].id
+
+This will have the effect of stopping the while loop when you eventually
+receive an empty list from ``craton.cells.list(...)``.
+
+Creating Cells
+--------------
+
+Cells live below a Region in Craton. To create a cell, the only required items
+are a ``name`` for the cell, a cloud ID, and a region ID. The name must be
+unique among cells in the same project.
+
+.. code-block:: python
+
+    cell = craton.cells.create(
+        name='my-cell-0',
+        cloud_id=cloud_id,
+        region_id=region_id,
+        note='This is my cell, there are many like it, but this is mine.',
+        variables={
+            'some-var': 'some-var-value',
+        },
+    )
+
+Retrieving a Specific Cell
+--------------------------
+
+Cells can be retrieved by id.
+
+.. code-block:: python
+
+    cell = craton.cells.get(1)
+
+Using a Cell's Variables
+-------------------------
+
+Once we have a cell we can introspect its variables like so:
+
+.. code-block:: python
+
+    cell = craton.cells.get(cell_id)
+    cell_vars = cell.variables.get()
+
+To update them:
+
+.. code-block:: python
+
+    updated_vars = {
+        'var-a': 'new-var-a',
+        'var-b': 'new-var-b',
+        'updated-var': 'updated value',
+    }
+    cell.variables.update(**updated_vars)
+
+To delete them:
+
+.. code-block:: python
+
+    cell.variables.delete('var-a', 'var-b', 'updated-var')
+
+Updating a Cell
+---------------
+
+We can update a cell's attributes (but not its variables) like so:
+
+.. code-block:: python
+
+    craton.cells.update(
+        cell_id,
+        name='new name',
+        note='Updated note.',
+    )
+
+Most attributes that you can specify on creation can also be specified for
+updating the cell as well.
+
+Deleting a Cell
+---------------
+
+We can delete with only its id:
+
+.. code-block:: python
+
+    craton.cells.delete(cell_id)
diff --git a/doc/source/usage/clouds.rst b/doc/source/usage/clouds.rst
new file mode 100644
index 0000000..f614b20
--- /dev/null
+++ b/doc/source/usage/clouds.rst
@@ -0,0 +1,134 @@
+======================
+ Using the Clouds API
+======================
+
+Here we will assume that we already have a
+:class:`~cratonclient.client.Client` instance configured with the appropriate
+authentication method (as demonstrated in :ref:`usage-auth`).
+
+Listing Clouds
+--------------
+
+The Clouds API implements pagination. This means that by default, it does not
+return all clouds known to Craton. To ignore page limits and offsets, we can
+allow cratonclient to do handle pagination for us:
+
+.. code-block:: python
+
+   for cloud in craton.clouds.list():
+       print_cloud_info(cloud)
+
+By default :meth:`~cratonclient.v1.clouds.CloudsManager.list` will handle
+pagination for you. If, instead, you want to handle it yourself you will want
+to do something akin to:
+
+.. code-block:: python
+
+    first_page_of_clouds = list(craton.clouds.list(autopaginate=False))
+    marker_id = first_page_of_clouds[-1].id
+    second_page_of_clouds = list(craton.clouds.list(
+        autopaginate=False,
+        marker=marker_id,
+    ))
+    marker_id = second_page_of_clouds[-1].id
+    third_page_of_clouds = list(craton.clouds.list(
+        autopaginate=False,
+        marker=marker_id,
+    ))
+    # etc.
+
+A more realistic example, however, might look like this:
+
+.. code-block:: python
+
+    clouds_list = None
+    marker = None
+    while clouds_list and clouds_list is not None:
+        clouds_list = list(craton.clouds.list(
+            marker=marker,
+            autopaginate=False,
+        ))
+        # do something with clouds_list
+        if clouds_list:
+            marker = clouds_list[-1].id
+
+This will have the effect of stopping the while loop when you eventually
+receive an empty list from ``craton.clouds.list(...)``.
+
+Creating Clouds
+---------------
+
+Clouds are the top-level item in Craton. To create a cloud, the only required
+item is a ``name`` for the cloud. This must be unique among clouds in the same
+project.
+
+.. code-block:: python
+
+    cloud = craton.clouds.create(
+        name='my-cloud-0',
+        note='This is my cloud, there are many like it, but this is mine.',
+        variables={
+            'some-var': 'some-var-value',
+        },
+    )
+
+Retrieving a Specific Cloud
+---------------------------
+
+Clouds can be retrieved by id.
+
+.. code-block:: python
+
+    cloud = craton.clouds.get(1)
+
+Using a Cloud's Variables
+-------------------------
+
+Once we have a cloud we can introspect its variables like so:
+
+.. code-block:: python
+
+    cloud = craton.clouds.get(cloud_id)
+    cloud_vars = cloud.variables.get()
+
+To update them:
+
+.. code-block:: python
+
+    updated_vars = {
+        'var-a': 'new-var-a',
+        'var-b': 'new-var-b',
+        'updated-var': 'updated value',
+    }
+    cloud.variables.update(**updated_vars)
+
+To delete them:
+
+.. code-block:: python
+
+    cloud.variables.delete('var-a', 'var-b', 'updated-var')
+
+Updating a Cloud
+----------------
+
+We can update a cloud's attributes (but not its variables) like so:
+
+.. code-block:: python
+
+    craton.clouds.update(
+        cloud_id,
+        name='new name',
+        note='Updated note.',
+    )
+
+Most attributes that you can specify on creation can also be specified for
+updating the cloud as well.
+
+Deleting a Cloud
+----------------
+
+We can delete with only its id:
+
+.. code-block:: python
+
+    craton.clouds.delete(cloud_id)
diff --git a/doc/source/usage/devices.rst b/doc/source/usage/devices.rst
new file mode 100644
index 0000000..7010aba
--- /dev/null
+++ b/doc/source/usage/devices.rst
@@ -0,0 +1,32 @@
+=======================
+ Using the Devices API
+=======================
+
+Here we will assume that we already have a
+:class:`~cratonclient.client.Client` instance configured with the appropriate
+authentication method (as demonstrated in :ref:`usage-auth`).
+
+.. note::
+
+    The Devices API is quite unlike other API endpoints presently in craton.
+    At the moment, it returns both hosts and network devices. It concatenates
+    the two lists in an indeterminate manner. On one invocation, you may
+    receive the hosts first and then the network devices, on another you may
+    receive them in the alternate order. If more items are returned in these
+    listings, then the number of different orderings will only increase
+    factorially.
+
+Listing Devices
+---------------
+
+The Devices API implements pagination. This means that by default, it does not
+return all devices known to Craton. To ignore page limits and offsets, we can
+allow cratonclient to do handle pagination for us:
+
+.. code-block:: python
+
+   for device in craton.devices.list():
+       print_device_info(device)
+
+By default :meth:`~cratonclient.v1.devices.DevicesManager.list` will handle
+pagination for you.
diff --git a/doc/source/usage/hosts.rst b/doc/source/usage/hosts.rst
new file mode 100644
index 0000000..304f79d
--- /dev/null
+++ b/doc/source/usage/hosts.rst
@@ -0,0 +1,148 @@
+=====================
+ Using the Hosts API
+=====================
+
+Here we will assume that we already have a
+:class:`~cratonclient.client.Client` instance configured with the appropriate
+authentication method (as demonstrated in :ref:`usage-auth`).
+
+Listing Hosts
+-------------
+
+The Hosts API implements pagination. This means that by default, it does not
+return all hosts known to Craton. To ignore page limits and offsets, we can
+allow cratonclient to do handle pagination for us:
+
+.. code-block:: python
+
+   for host in craton.hosts.list():
+       print_host_info(host)
+
+By default :meth:`~cratonclient.v1.hosts.HostsManager.list` will handle
+pagination for you. If, instead, you want to handle it yourself you will want
+to do something akin to:
+
+.. code-block:: python
+
+    first_page_of_hosts = list(craton.hosts.list(autopaginate=False))
+    marker_id = first_page_of_hosts[-1].id
+    second_page_of_hosts = list(craton.hosts.list(
+        autopaginate=False,
+        marker=marker_id,
+    ))
+    marker_id = second_page_of_hosts[-1].id
+    third_page_of_hosts = list(craton.hosts.list(
+        autopaginate=False,
+        marker=marker_id,
+    ))
+    # etc.
+
+A more realistic example, however, might look like this:
+
+.. code-block:: python
+
+    hosts_list = None
+    marker = None
+    while hosts_list and hosts_list is not None:
+        hosts_list = list(craton.hosts.list(
+            marker=marker,
+            autopaginate=False,
+        ))
+        # do something with hosts_list
+        if hosts_list:
+            marker = hosts_list[-1].id
+
+This will have the effect of stopping the while loop when you eventually
+receive an empty list from ``craton.hosts.list(...)``.
+
+Creating Hosts
+--------------
+
+Hosts live inside either a Region or Cell in Craton. To create a host, one
+needs:
+
+- A unique name
+
+- A unique IP address
+
+- A "device type" (this is freeform), e.g., "server", "container", "nova-vm",
+  etc.
+
+- A cloud ID
+
+- A region ID
+
+.. code-block:: python
+
+    host = craton.hosts.create(
+        name='my-host-0',
+        ip_address='127.0.1.0',
+        device_type='server',
+        cloud_id=cloud_id,
+        region_id=region_id,
+        note='This is my host, there are many like it, but this is mine.',
+        variables={
+            'some-var': 'some-var-value',
+        },
+    )
+
+Retrieving a Specific Host
+--------------------------
+
+Hosts can be retrieved by id.
+
+.. code-block:: python
+
+    host = craton.hosts.get(1)
+
+Using a Host's Variables
+------------------------
+
+Once we have a host we can introspect its variables like so:
+
+.. code-block:: python
+
+    host = craton.hosts.get(host_id)
+    host_vars = host.variables.get()
+
+To update them:
+
+.. code-block:: python
+
+    updated_vars = {
+        'var-a': 'new-var-a',
+        'var-b': 'new-var-b',
+        'updated-var': 'updated value',
+    }
+    host.variables.update(**updated_vars)
+
+To delete them:
+
+.. code-block:: python
+
+    host.variables.delete('var-a', 'var-b', 'updated-var')
+
+Updating a Host
+---------------
+
+We can update a host's attributes (but not its variables) like so:
+
+.. code-block:: python
+
+    craton.hosts.update(
+        host_id,
+        name='new name',
+        note='Updated note.',
+    )
+
+Most attributes that you can specify on creation can also be specified for
+updating the host as well.
+
+Deleting a Host
+---------------
+
+We can delete with only its id:
+
+.. code-block:: python
+
+    craton.hosts.delete(host_id)
diff --git a/doc/source/usage/index.rst b/doc/source/usage/index.rst
new file mode 100644
index 0000000..918e023
--- /dev/null
+++ b/doc/source/usage/index.rst
@@ -0,0 +1,30 @@
+=======================
+ Python API User Guide
+=======================
+
+Once you have installed ``python-cratonclient``, there are a few things you
+need to get started using the Python API:
+
+#. You need to know how to authenticate to the Craton API you wish to talk to
+
+   Some Craton API services will be deployed using Craton's in-built
+   authentication system while others may use Keystone.
+
+#. You need your credentials
+
+#. You need the location of your Craton API service
+
+This chapter of python-cratonclient's documentation is broken down into
+chapters:
+
+.. toctree::
+    :maxdepth: 2
+
+    authentication
+    supported-apis
+    clouds
+    regions
+    cells
+    hosts
+    devices
+    projects
diff --git a/doc/source/usage/projects.rst b/doc/source/usage/projects.rst
new file mode 100644
index 0000000..7de0a14
--- /dev/null
+++ b/doc/source/usage/projects.rst
@@ -0,0 +1,135 @@
+========================
+ Using the Projects API
+========================
+
+Here we will assume that we already have a
+:class:`~cratonclient.client.Client` instance configured with the appropriate
+authentication method (as demonstrated in :ref:`usage-auth`).
+
+Listing Projects
+----------------
+
+The Projects API implements pagination. This means that by default, it does not
+return all projects known to Craton. To ignore page limits and offsets, we can
+allow cratonclient to do handle pagination for us:
+
+.. code-block:: python
+
+   for project in craton.projects.list():
+       print_project_info(project)
+
+By default :meth:`~cratonclient.v1.projects.ProjectsManager.list` will handle
+pagination for you. If, instead, you want to handle it yourself you will want
+to do something akin to:
+
+.. code-block:: python
+
+    first_page_of_projects = list(craton.projects.list(autopaginate=False))
+    marker_id = first_page_of_projects[-1].id
+    second_page_of_projects = list(craton.projects.list(
+        autopaginate=False,
+        marker=marker_id,
+    ))
+    marker_id = second_page_of_projects[-1].id
+    third_page_of_projects = list(craton.projects.list(
+        autopaginate=False,
+        marker=marker_id,
+    ))
+    # etc.
+
+A more realistic example, however, might look like this:
+
+.. code-block:: python
+
+    projects_list = None
+    marker = None
+    while projects_list and projects_list is not None:
+        projects_list = list(craton.projects.list(
+            marker=marker,
+            autopaginate=False,
+        ))
+        # do something with projects_list
+        if projects_list:
+            marker = projects_list[-1].id
+
+This will have the effect of stopping the while loop when you eventually
+receive an empty list from ``craton.projects.list(...)``.
+
+Creating Projects
+-----------------
+
+Projects are top-level items in Craton. To create a project, one needs:
+
+- A unique name
+
+- Permission to create new projects
+
+
+.. code-block:: python
+
+    project = craton.projects.create(
+        name='my-project-0',
+        variables={
+            'some-var': 'some-var-value',
+        },
+    )
+
+Retrieving a Specific Project
+-----------------------------
+
+Projects can be retrieved by id.
+
+.. code-block:: python
+
+    project = craton.projects.get(1)
+
+Using a Project's Variables
+---------------------------
+
+Once we have a project we can introspect its variables like so:
+
+.. code-block:: python
+
+    project = craton.projects.get(project_id)
+    project_vars = project.variables.get()
+
+To update them:
+
+.. code-block:: python
+
+    updated_vars = {
+        'var-a': 'new-var-a',
+        'var-b': 'new-var-b',
+        'updated-var': 'updated value',
+    }
+    project.variables.update(**updated_vars)
+
+To delete them:
+
+.. code-block:: python
+
+    project.variables.delete('var-a', 'var-b', 'updated-var')
+
+Updating a Project
+------------------
+
+We can update a project's attributes (but not its variables) like so:
+
+.. code-block:: python
+
+    craton.projects.update(
+        project_id,
+        name='new name',
+    )
+
+Most attributes that you can specify on creation can also be specified for
+updating the project as well.
+
+Deleting a Project
+------------------
+
+We can delete with only its id:
+
+.. code-block:: python
+
+    craton.projects.delete(project_id)
diff --git a/doc/source/usage/regions.rst b/doc/source/usage/regions.rst
new file mode 100644
index 0000000..2f487b8
--- /dev/null
+++ b/doc/source/usage/regions.rst
@@ -0,0 +1,135 @@
+=======================
+ Using the Regions API
+=======================
+
+Here we will assume that we already have a
+:class:`~cratonclient.client.Client` instance configured with the appropriate
+authentication method (as demonstrated in :ref:`usage-auth`).
+
+Listing Regions
+---------------
+
+The Regions API implements pagination. This means that by default, it does not
+return all regions known to Craton. To ignore page limits and offsets, we can
+allow cratonclient to do handle pagination for us:
+
+.. code-block:: python
+
+   for region in craton.regions.list():
+       print_region_info(region)
+
+By default :meth:`~cratonclient.v1.regions.RegionsManager.list` will handle
+pagination for you. If, instead, you want to handle it yourself you will want
+to do something akin to:
+
+.. code-block:: python
+
+    first_page_of_regions = list(craton.regions.list(autopaginate=False))
+    marker_id = first_page_of_regions[-1].id
+    second_page_of_regions = list(craton.regions.list(
+        autopaginate=False,
+        marker=marker_id,
+    ))
+    marker_id = second_page_of_regions[-1].id
+    third_page_of_regions = list(craton.regions.list(
+        autopaginate=False,
+        marker=marker_id,
+    ))
+    # etc.
+
+A more realistic example, however, might look like this:
+
+.. code-block:: python
+
+    regions_list = None
+    marker = None
+    while regions_list and regions_list is not None:
+        regions_list = list(craton.regions.list(
+            marker=marker,
+            autopaginate=False,
+        ))
+        # do something with regions_list
+        if regions_list:
+            marker = regions_list[-1].id
+
+This will have the effect of stopping the while loop when you eventually
+receive an empty list from ``craton.regions.list(...)``.
+
+Creating Regions
+----------------
+
+Regions are required to be part of a Cloud in Craton. To create a region, the
+only required items are a ``name`` for the region and the ID of the cloud it
+belongs to. The name must be unique among regions in the same project.
+
+.. code-block:: python
+
+    region = craton.regions.create(
+        name='my-region-0',
+        cloud_id=cloud_id,
+        note='This is my region, there are many like it, but this is mine.',
+        variables={
+            'some-var': 'some-var-value',
+        },
+    )
+
+Retrieving a Specific Region
+----------------------------
+
+Regions can be retrieved by id.
+
+.. code-block:: python
+
+    region = craton.regions.get(1)
+
+Using a Region's Variables
+--------------------------
+
+Once we have a region we can introspect its variables like so:
+
+.. code-block:: python
+
+    region = craton.regions.get(region_id)
+    region_vars = region.variables.get()
+
+To update them:
+
+.. code-block:: python
+
+    updated_vars = {
+        'var-a': 'new-var-a',
+        'var-b': 'new-var-b',
+        'updated-var': 'updated value',
+    }
+    region.variables.update(**updated_vars)
+
+To delete them:
+
+.. code-block:: python
+
+    region.variables.delete('var-a', 'var-b', 'updated-var')
+
+Updating a Region
+-----------------
+
+We can update a region's attributes (but not its variables) like so:
+
+.. code-block:: python
+
+    craton.regions.update(
+        region_id,
+        name='new name',
+        note='Updated note.',
+    )
+
+Most attributes that you can specify on creation can also be specified for
+updating the region as well.
+
+Deleting a Region
+-----------------
+
+We can delete with only its id:
+
+.. code-block:: python
+
+    craton.regions.delete(region_id)
diff --git a/doc/source/usage/supported-apis.rst b/doc/source/usage/supported-apis.rst
new file mode 100644
index 0000000..eb19f19
--- /dev/null
+++ b/doc/source/usage/supported-apis.rst
@@ -0,0 +1,50 @@
+===========================
+ Communicating with Craton
+===========================
+
+Now that you've configured your authentication method, you can interact with
+your ``craton`` object like so:
+
+.. code-block:: python
+
+    for region in craton.regions.list():
+        print('Region {} contains:'.format(region.name))
+        for host in craton.hosts.list(region_id=region.id):
+            print('    {}'.format(host.name))
+
+
+The Craton API has the following resources:
+
+- Cells
+
+- Clouds
+
+- Devices
+
+- Hosts
+
+- Network Devices
+
+- Network Interfaces
+
+- Networks
+
+- Projects
+
+- Regions
+
+- Users
+
+Of these:
+
+- Cells
+
+- Clouds
+
+- Hosts
+
+- Projects
+
+- Regions
+
+Are implemented.