diff --git a/.env.example b/.env.example
new file mode 100644
index 0000000..27f464f
--- /dev/null
+++ b/.env.example
@@ -0,0 +1,4 @@
+OPENAI_API_KEY=sk-proj-...
+
+RAPIDAPI_KEY=9df2cb5...
+RAPIDAPI_HOST=sky-scrapper.p.rapidapi.com
\ No newline at end of file
diff --git a/README.md b/README.md
index 4a558ad..46b2575 100644
--- a/README.md
+++ b/README.md
@@ -5,10 +5,17 @@ Work in progress.
 This demo shows a multi-turn conversation with an AI agent running inside a Temporal workflow. The goal is to collect information towards a goal. There's a simple DSL input for collecting information (currently set up to use mock functions to search for events, book flights around those events then generate an invoice for those flights, see `send_message.py`). The AI will respond with clarifications and ask for any missing information to that goal. It uses a local LLM via Ollama.
 
 ## Setup
-* Install [Ollama](https://ollama.com) and the [Qwen2.5 14B](https://ollama.com/library/qwen2.5) model (`ollama run qwen2.5:14b`). (note this model is about 9GB to download).
+* Requires an OpenAI key for the gpt-4o model. Set this in the `OPENAI_API_KEY` environment variable in .env
+* Requires a rapidapi key for sky-scrapper (how we find flights). Set this in the `RAPIDAPI_KEY` environment variable in .env
+    * If you're lazy go to `tools/search_flights.py` and replace the `get_flights` function with the mock `search_flights_example` that exists in the same file.
+* See .env_example for the required environment variables.
 * Install and run Temporal. Follow the instructions in the [Temporal documentation](https://learn.temporal.io/getting_started/python/dev_environment/#set-up-a-local-temporal-service-for-development-with-temporal-cli) to install and run the Temporal server.
 * Install the dependencies: `poetry install`
 
+Deprecated:
+* Install [Ollama](https://ollama.com) and the [Qwen2.5 14B](https://ollama.com/library/qwen2.5) model (`ollama run qwen2.5:14b`). (note this model is about 9GB to download).
+  * Local LLM is disabled as ChatGPT 4o was better for this use case. To use Ollama, examine `./activities/tool_activities.py` and rename the functions.
+
 ## Running the example
 
 From the /scripts directory: