diff --git a/README.md b/README.md
index 65d5d22..d479ddf 100644
--- a/README.md
+++ b/README.md
@@ -23,8 +23,8 @@ features, but higher quality in the output. Check out the details at https://git
 
 There are three different ways to get started with SP. Each has its own document:
 
- * [Server](server/README.md): This contains the SP server. It's useful for PDF parsing as a service. It's also probably the easiest way to get going.
  * [CLI](cli/README.md): This contains the command line interface to SP. That's most useful for batch processing.
+ * [Server](server/README.md): This contains the SP server. It's useful for PDF parsing as a service.
  * [Core](core/README.md): This contains SP as a library. It has all the extraction code, plus training and evaluation. Both server and CLI use this to do the actual work.
 
 ## How to include into your own project
diff --git a/cli/README.md b/cli/README.md
index 84788fa..bcbd925 100644
--- a/cli/README.md
+++ b/cli/README.md
@@ -1,14 +1,13 @@
 # Science Parse Command Line Interface
 
-Science Parse has a command line interface called "RunSP". To build it into a super-jar, run `sbt cli/assembly`. This will download all dependencies and make a bundled jar that contains SP itself, plus all of its dependencies.
+Download the latest science-parse-cli-assembly executable from the [Releases](https://github.com/allenai/science-parse/releases) page
 
 You can run it like this:
 ```
 java -jar jarfile.jar --help
 ```
-That will print all the command line options you can use with it. I won't describe all of them here, but there are a few important described below.
 
-## Memory
+That will print all the command line options you can use with it. I won't describe all of them here, but there are a few important described below.
 
 Science Parse needs quite a lot of memory. We recommend you run it with at least 6GB of heap, like this:
 ```
@@ -16,6 +15,10 @@ java -Xmx6g -jar jarfile.jar 18bc3569da037a6cb81fb081e2856b77b321c139
 ```
 Note that some documents need more memory to parse than others.
 
+## Building from source
+
+Science Parse has a command line interface called "RunSP". To build it into a super-jar, run `sbt cli/assembly`. This will download all dependencies and make a bundled jar that contains SP itself, plus all of its dependencies.
+
 ## Specifying input
 
 `RunSP` can parse multiple files at the same time. You can parse thousands of PDFs like this. It will try to parse as many of them in parallel as your computer allows.