diff --git a/.github/workflows/trampoline.yaml b/.github/workflows/trampoline.yaml
index efdd0a452..0b320c7c1 100644
--- a/.github/workflows/trampoline.yaml
+++ b/.github/workflows/trampoline.yaml
@@ -85,6 +85,12 @@ jobs:
           mkdir -p trampolines
           # Iterate through all files in trampolines directory and its subdirectories
           find trampolines-binaries -type f  -name 'pixi-trampoline-*' -exec mv -f {} trampolines/ \;
+          # now iterate through all files in trampolines directory and compress them using zstd
+          # and remove the original file
+          for file in trampolines/*; do
+            zstd "$file"
+            rm "$file"
+          done
           ls -R trampolines/
 
       - name: Upload binary artifact
diff --git a/crates/pixi_trampoline/README.md b/crates/pixi_trampoline/README.md
index 014099ebe..08a480d05 100644
--- a/crates/pixi_trampoline/README.md
+++ b/crates/pixi_trampoline/README.md
@@ -18,5 +18,6 @@ This is the configuration used by trampoline to set the env variables, and run t
 
 # How to build it?
 You can use `trampoline.yaml` workflow to build the binary for all the platforms and architectures supported by pixi.
-If running it manually from forked repo, you can download and commit artifacts to your branch.
+In case of building it manually, you can use the following command, after executing the `cargo build --release`, you need to compress it using `zstd`.
+Then you can commit `compressed` artifacts to your branch.
 If running it manually or triggered by changes in `crates/pixi_trampoline` from the main repo, they will be automatically committed to the branch.