diff --git a/assistant/assistant.py b/assistant/assistant.py
index 05568648..f86f799a 100644
--- a/assistant/assistant.py
+++ b/assistant/assistant.py
@@ -23,16 +23,13 @@
 class LLM:
     def __init__(self, chat_model: str):
         self.chat_model = chat_model
-        self.client = OpenAI(
-            # This is the default and can be omitted
-            api_key=os.environ.get("OPENAI_API_KEY"),
-        )
+        self.llm_client = OpenAI()
         self.messages = []
         self.append_message(ROLE_SYSTEM, os.getenv("ASSISTANT_ROLE"))
 
     def chat(self, content: str) -> dict:
         self.append_message(ROLE_USER, content)
-        response = self.client.chat.completions.create(
+        response = self.llm_client.chat.completions.create(
             model=self.chat_model, messages=self.messages
         )
         logging.info(f"OPENAI RESPONSE: {response.choices[0].message.content}")
@@ -41,7 +38,7 @@ def chat(self, content: str) -> dict:
         return message.content
 
     def append_message(self, role, content):
-        self.messages.append({"role": role, "content": str(content)})
+        self.messages.append({MSG_ROLE: role, MSG_CONTENT: str(content)})
 
 
 llm = LLM(os.getenv("OPENAI_CHAT_MODEL"))
diff --git a/readme.md b/readme.md
index 2b62be6a..b4404158 100644
--- a/readme.md
+++ b/readme.md
@@ -39,18 +39,17 @@ retrieval part of the system.
 | `LIBRARIAN_COLLECTION` | knowledge-base | The name of the ChromaDB collection where the corpus will be stored. |
 
 
-## Ollama LLM Runtime
+## OpenAI LLM Runtime
 
-We use [Ollama](https://ollama.com/) as the model run-time. Ollama makes it easy
-to manage multiple models, without having to handle large model files. It also
-solves the need for registration on all kinds of sites. Ollama will just pull
-the model in and cache it locally.
+We use [OpenAI](https://platform.openai.com/) as the model run-time. OpenAI provides robust capabilities for managing multiple models and handling large model files. It simplifies the integration process by managing registrations and pulling models as needed.
 
 | `.env` | default | description |
 |---|---|---|
-| `OLLAMA_HOST` | _CHANGEME_ | The IP address or DNS name of the host that runs Ollama. |
-| `OLLAMA_PORT` | 11434 | The port number on the host that runs Ollama. |
-| `OLLAMA_CHAT_MODEL` | mistral | The LLM model that is used to handle chat messages. |
+| `OPENAI_API_KEY` | _CHANGEME_ | The API key for authenticating with OpenAI services. |
+| `OPENAI_CHAT_MODEL` | `gpt-3.5` / `gpt-4` | The LLM model that is used to handle chat messages. Read more about [OpenAI models](https://platform.openai.com/docs/models) |
+
+Note: The __OPENAI_API_KEY__ does not need to be explicitly called in [assistant.py](https://github.com/akvo/rag-doll/blob/master/assistant/assistant.py) because the openai library automatically reads it from the environment variables when `openai.OpenAI()` is instantiated.
+
 
 
 ## ChromaDB Vector Database