diff --git a/.gitignore b/.gitignore
index b6e4761..855b8d2 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,3 +1,5 @@
+_temp.py
+
 # Byte-compiled / optimized / DLL files
 __pycache__/
 *.py[cod]
diff --git a/README.md b/README.md
index d875e37..a3307b2 100644
--- a/README.md
+++ b/README.md
@@ -1,2 +1,24 @@
-# violet
+# ***V i o l e t G P T***
 🤖 A next-generation chat bot based on OpenAI, which can access your files, open the webbrowser and more!
+
+## Features
+- Accurate mathematical calculations
+- File system access
+- Voice support (SpeechToText + TextToSpeech)
+
+### Planned
+- API access (save your API tokens for various services in a `.env` file, and the bot will be able to find out e.g. the weather.)
+- Web UI planned (the sandbox has a way higher priority, though! No web UI is going to be worked on before there is no proper sandbox.)
+
+## Warning
+The sandbox is currently being developed, which means, in theory, the bot do anything.
+
+Never, ever run the bot with administrator permissions. You should always make sure the sandbox can't delete any imporant files!
+Additionally, never ever, give others access to the sandbox features! Malicious actors could use this as a remote code execution.
+
+## Support
+✅ Debian-Based Linux Distros (such as Mint 20.2)
+
+✅ Windows 10 (and above)
+
+✅ MacOS (probably - testers needed!)
\ No newline at end of file
diff --git a/requirements.txt b/requirements.txt
new file mode 100644
index 0000000..3f4e1ba
--- /dev/null
+++ b/requirements.txt
@@ -0,0 +1,2 @@
+openai
+colorama
\ No newline at end of file
diff --git a/run.py b/run.py
new file mode 100644
index 0000000..f34fd20
--- /dev/null
+++ b/run.py
@@ -0,0 +1,20 @@
+import sys
+import colorama
+
+import violet
+
+colorama.init(autoreset=True)
+
+if len(sys.argv) > 1:
+    prompt = ' '.join(sys.argv[1:])
+    print(violet.respond(prompt)['colored'])
+
+else:
+    while True:
+        try:
+            prompt = input(colorama.Fore.BLUE)
+        except KeyboardInterrupt:
+            sys.exit(0)
+
+        print(colorama.Fore.RESET, end='')
+        print(violet.respond(prompt)['colored'])
diff --git a/violet/__init__.py b/violet/__init__.py
new file mode 100644
index 0000000..2afa198
--- /dev/null
+++ b/violet/__init__.py
@@ -0,0 +1,4 @@
+from .gpt import *
+from .main import *
+from .helpers import *
+from .sandbox import *
diff --git a/violet/error-model.txt b/violet/error-model.txt
new file mode 100644
index 0000000..34f939e
--- /dev/null
+++ b/violet/error-model.txt
@@ -0,0 +1,27 @@
+The following is a conversation between a Human and an AI which explains Python errors.
+The AI gives instructions on how the issue could be solved, if necessary.
+
+Human: ```
+Traceback (most recent call last):
+  File "sandbox/_temp.py", line 3, in <module>
+    subprocess.call(["nautilus"])
+  File "/usr/lib/python3.8/subprocess.py", line 340, in call
+    with Popen(*popenargs, **kwargs) as p:
+  File "/usr/lib/python3.8/subprocess.py", line 858, in __init__
+    self._execute_child(args, executable, preexec_fn, close_fds,
+  File "/usr/lib/python3.8/subprocess.py", line 1704, in _execute_child
+    raise child_exception_type(errno_num, err_msg, err_filename)
+FileNotFoundError: [Errno 2] No such file or directory: 'nautilus'
+```
+AI: Sorry, you the process "nautilus" could not be found. Please try another command.
+Human: ```
+Traceback (most recent call last):
+  File "run.py", line 20, in <module>
+    print(violet.respond(prompt)['colored'])
+  File "/home/onlix/Documents/NSDE/violet/violet/main.py", line 18, in respond
+    response = sandbox.run(code)        
+  File "/home/onlix/Documents/NSDE/violet/violet/sandbox.py", line 22, in run
+    return output
+NameError: name 'output' is not defined
+```
+AI: I'm sorry, but the bot was unable to respond. Please try asking a different question.
\ No newline at end of file
diff --git a/violet/gpt.py b/violet/gpt.py
new file mode 100644
index 0000000..413f1fb
--- /dev/null
+++ b/violet/gpt.py
@@ -0,0 +1,39 @@
+import os
+import openai
+
+from dotenv import load_dotenv
+
+load_dotenv()
+
+openai.api_key = os.getenv('OPENAI_API_KEY') 
+error_model = open('violet/error-model.txt').read()
+
+def generate(history: str, prompt: str) -> str:
+    completions = openai.Completion.create(
+        model="text-davinci-003",
+        prompt=f'{history}\nHuman: {prompt}\nAI: ',
+        temperature=0.9,
+        max_tokens=300,
+        top_p=1,
+        frequency_penalty=0.0,
+        presence_penalty=0.6,
+        stop=[" Human:", " AI:"]
+    )
+
+    message = completions.choices[0].text
+    return message[1:]
+
+def explain_error(prompt: str) -> str:
+    completions = openai.Completion.create(
+        model="text-davinci-003",
+        prompt=f'{error_model}\nHuman: {prompt}\nAI: ',
+        temperature=0.9,
+        max_tokens=300,
+        top_p=1,
+        frequency_penalty=0.0,
+        presence_penalty=0.6,
+        stop=[" Human:", " AI:"]
+    )
+
+    message = completions.choices[0].text
+    return message[1:]
diff --git a/violet/helpers.py b/violet/helpers.py
new file mode 100644
index 0000000..890bc67
--- /dev/null
+++ b/violet/helpers.py
@@ -0,0 +1,3 @@
+import platform
+
+opsys = f"{platform.system().replace('Darwin', 'MacOS')} {platform.machine()}"
diff --git a/violet/main.py b/violet/main.py
new file mode 100644
index 0000000..0c47ecf
--- /dev/null
+++ b/violet/main.py
@@ -0,0 +1,33 @@
+import colorama
+
+from . import gpt
+from . import helpers
+from . import sandbox
+
+CODE_START = '```\n'
+CODE_END = '\n```'
+
+history = open('violet/model.txt').read().replace('#OS#', helpers.opsys)
+
+def respond(prompt: str) -> str:
+    global history
+    response = gpt.generate(history=history, prompt=prompt)
+    
+    if CODE_START in response and CODE_END in response:
+        code = response.split(CODE_START)[1].split(CODE_END)[0]
+        response = sandbox.run(code)       
+        colored = colorama.Fore.GREEN if response['status'] == 'success' else colorama.Fore.RED + response['message']
+    else:
+        colored = colorama.Fore.YELLOW + response
+
+    history += \
+f"""
+Human: {prompt}
+AI: {response}
+"""
+
+    return {
+        'plain': response,
+        'colored': colored 
+    }
+
diff --git a/violet/model.txt b/violet/model.txt
new file mode 100644
index 0000000..7d2d565
--- /dev/null
+++ b/violet/model.txt
@@ -0,0 +1,42 @@
+The following is a conversation with an AI assistant.
+The assistant is helpful, creative, clever, and smart.
+For some specific questions, such as time, mathematical equations and more, the assistant writes Python code which is then being ran and the output is being displayed.
+The code is being ran on #OS#, so all code blocks are made to work on #OS#.
+
+Human: Hello, who are you?
+AI: I am an AI created by OpenAI. How can I help you today?
+Human: What's the current time?
+AI: ```
+import time
+
+current_time = time.localtime()
+current_time = time.strftime("%H:%M:%S", current_time)
+
+print(f"The current time is {current_time}.")
+```
+Human: Thanks. Please create a file called "todo.txt" and save "Read a book" in it."
+AI: ```
+with open("todo.txt", "w") as f:
+    f.write("Read a book")
+print("Done, the file has been written.".)
+```
+Human: What is the capital of France?
+AI: Paris.
+Human: Please read my todo.txt.
+AI: ```
+with open("todo.txt", "r", encoding="utf8") as f:
+    print("The content of todo.txt is: " + f.read())
+```
+Human: Thanks!
+AI: You're welcome.
+Human: Now, please open YouTube.
+AI: ```
+import webbrowser
+
+webbrowser.open("https://youtube.com")
+print("Done.")
+```
+Human: What's 3 to the power of 4 plus 6 multiplied by 5?
+AI: ```
+print(f"{3**4+6*5}")
+```
\ No newline at end of file
diff --git a/violet/sandbox.py b/violet/sandbox.py
new file mode 100644
index 0000000..dc883ef
--- /dev/null
+++ b/violet/sandbox.py
@@ -0,0 +1,27 @@
+import colorama
+import subprocess
+
+from . import gpt
+
+def run(code: str):
+    print(colorama.Style.DIM + 'Launching sandbox...')
+    with open('sandbox/_temp.py', 'w') as f:
+        f.write(code)
+
+    # output = subprocess.check_output(, stderr=subprocess.DEVNULL).decode('utf8').strip()
+    process = subprocess.Popen(['python', 'sandbox/_temp.py'], stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+    stdout, stderr = process.communicate()
+
+    output = stdout.decode('utf8').strip()
+    errors = stderr.decode('utf8').strip()
+
+    if errors:
+        return {
+        'status': 'error',
+        'message': gpt.explain_error(errors)
+    }
+
+    return {
+        'status': 'success',
+        'message': output
+    }
diff --git a/voice.py b/voice.py
new file mode 100644
index 0000000..85b08be
--- /dev/null
+++ b/voice.py
@@ -0,0 +1,21 @@
+import sys
+import colorama
+import pyttsx3
+
+import violet
+
+colorama.init(autoreset=True)
+engine = pyttsx3.init()
+
+while True:
+    try:
+        prompt = input(colorama.Fore.BLUE)
+
+    except KeyboardInterrupt:
+        sys.exit(0)
+
+    answer = violet.respond(prompt)
+    print(answer['colored'])
+    
+    engine.say(answer['plain'])
+    engine.runAndWait()