diff --git a/everythingapi/__init__.py b/everythingapi/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/everythingapi/__main__.py b/everythingapi/__main__.py
new file mode 100644
index 0000000..44b2bc1
--- /dev/null
+++ b/everythingapi/__main__.py
@@ -0,0 +1,104 @@
+from __future__ import annotations
+
+import json
+import os
+
+from dotenv import load_dotenv
+from flask import Flask, request, Response
+from openai import OpenAI
+from groq import Groq
+
+SYSTEM_PROMPT = """
+    You are the "Everything API" that can deal with all sorts of data and operations.
+    Answer all requests with some valid JSON that fits the given request.
+    Make sure to only return valid JSON. The response should not contain any comments or ellipsis.
+
+    For example: Assume you get a request for /cars/123 with method GET, your response could look like this:
+    {
+        "id": 123,
+        "make": "Toyota",
+        "model": "Camry",
+        "year": 2021,
+        "color": "Silver",
+        "mileage": 15000,
+        "engine": {
+            "type": "V6",
+            "horsepower": 301
+        },
+        "transmission": "Automatic",
+        "features": ["Leather seats", "Sunroof", "Navigation system"],
+        "price": 25000
+    }
+    """
+USER_PROMPT = """
+    URI: {uri}
+    Method: {method}
+    Request Body:
+    {body}
+    """
+
+ai: OpenAI | Groq | None = None
+
+app = Flask(__name__)
+
+
+@app.route("/")
+def root():
+    return Response(
+        '{"motd": "Welcome to the Everything API. Please provide an endpoint."}',
+        mimetype='application/json'
+    )
+
+
+@app.route("/favicon.ico")
+def favicon():
+    return Response('{"error": "not found"}', status=404, mimetype='application/json')
+
+
+@app.route("/<path:path>")
+def handler(path):
+    messages = [
+        {
+            "role": "system",
+            "content": SYSTEM_PROMPT
+        },
+        {
+            "role": "user",
+            "content": USER_PROMPT.format(
+                uri=request.url,
+                method=request.method,
+                body=request.data[:500]
+            )
+        }
+    ]
+
+    print("request:", request.method, request.url, request.data)
+
+    for i in range(0, 5):
+        try:
+            response = ai.chat.completions.create(
+                # model="gpt-3.5-turbo",
+                model="llama3-70b-8192",
+                messages=messages
+            ).choices[0].message.content
+            json.loads(response)  # verify response
+
+            print("response: " + response)
+            return Response(response, status=200, mimetype="application/json")
+        except ValueError:
+            print("response not parseable: " + response)
+            continue
+
+    return Response('{"error": "internal error"}', status=500, mimetype='application/json')
+
+
+def main():
+    global ai
+    load_dotenv()
+    # ai = OpenAI(api_key=os.environ["OPENAI_TOKEN"])
+    ai = Groq(api_key=os.environ["GROQ_TOKEN"])
+    app.run()
+
+
+if __name__ == "__main__":
+    main()