😺 Your own desktop GPT

Welcome, humans.

Happy holiday weekend to everyone in the U.S. (and to those international readers, hopefully you had fun too)!

For today’s starters: ever wondered what an 80s movie version of Super Smash Bros melee would look like? No? Well, now you know!

All the non-human characters are puppets, and honestly, we’re OBSESSED.

Imagine playing a 1980s version of Super Smash and have it look exactly like that video. In a few years, after all these data centers get built out and the cost of compute for 3D GFX goes way down, it’ll definitely be possible.

You can bet there’ll be some effort to create reskins not just for characters, but entire games. You better believe it’ll be a whole DLC, though. Game pubs gotta make that $$.

Advertise in The Neuron here.

How to build your own DIY chatbot tool.

Example custom chatbot in action.

If you're still copying and pasting all your prompts back and forth between ChatGPT's and Claude’s web interface, spoiler alert: you're doing AI wrong.

Today, we're going to use AI to help us build something way cooler: your own personal chatbot interface that can switch between ChatGPT and Claude with a single click.

How? We'll use AI to help us write the code, of course!

Step 1: Get some API keys.

First things first:

1. Get an OpenAI API key from platform.openai.com/api-keys.
2. Get a Claude API key from console.anthropic.com/settings/keys.

Store these somewhere safe—we'll need them in a minute.

IMPORTANT: before you’re able to use these API keys, you’ll need to load up some credits on your account. For billing and pricing details, see here for OpenAI and here for Claude. If you have a specific model you want to use, save that for the next step.

Step 2: Ask AI to Help Us Code

Here's where it gets fun. We're now going to use ChatGPT or Claude to help us build this thing.

1. Fire up your favorite AI (Claude or ChatGPT—doesn’t matter which, just pick the smartest version available).
2. Copy the quick start details (for Claude here and for ChatGPT here) and paste them into your chat for context.
3. Now, ask your AI to help you create a simple interface to access both APIs from your computer.
   1. To use a specific model in your Chat interface, make sure to specify the right model name (ChatGPT models, Claude models).

Here's the prompt we used:

Please help me create a Python script that builds a GUI application using PyQt6 to serve as an AI API switcher. The application should have the following features:

1. A dialog to securely store the API keys for both OpenAI and Anthropic using the keyring library.

2. A chat window with a dropdown to select between the OpenAI (ChatGPT) and Anthropic (Claude) APIs.

3. A chat display area to show the user's messages and the AI's responses.

4. An input field and send button to allow the user to send messages to the selected AI.

5. Error handling to display appropriate error messages if the API calls fail.

6. The application should make the API calls using the requests library and display the responses in the chat window.

Please provide the complete Python script with comments explaining the key steps.

Additionally, provide the steps to create a standalone .exe executable from the Python code that can be easily run on an end-user machine without requiring the terminal to start and stop the application (and handle security of API keys for this solution accordingly).

Use model name [enter model name] for Claude and [model name] for OpenAI.

^ Feel free to customize the features to your needs.

Step 3: Deploy Your Chatbot

Now follow the steps that your AI provides to install the code and make it a usable application, and that's it! You’ll have a working chatbot tool that can switch between both Claude and ChatGPT in the same window. Pretty cool, right?

NOTE: If you get stuck, ask your AI to “explain this to me like an amateur and walk me through it step by step from the beginning.”

What do you want to learn next?

How to train my own model.

How to run a model locally on my machine.

How to build an AI agent.

How to build upon this base chatbot with more features.

How to use another tool (please specify which tool).

FROM OUR PARTNERS

Want to build AI projects faster and cheaper?

Want to build AI projects faster and cheaper? Join Speed Read AI's founders and Dell's experts on Dec 12th to discover how to run AI locally with Precision workstations and NVIDIA RTX™.

Register here.

Around the Horn.

Keep in mind, this is probably being teleoperated, but the hand itself is CREEPY human-like.

• A bunch of Canadian publishers (CBC, Postmedia, Torstar) are suing OpenAI over copyright and terms of use breach. Interesting timing, because a new study found ChatGPT can only cite its sources accurately some of the time.
• Also, here’s a good explainer on where one of the major publisher copyright lawsuits against the company stands today. And just for fun, related: “The AI reporter who took my job got fired.”

Treats To Try.

1. *Learn how to run AI locally with Precision workstations and NVIDIA RTX™ from Speed Read AI's founders and Dell's AI expert. Register here.
2. Magicroll creates videos in one click with motion graphics and B-roll footage.
3. OfferGenie is an interview copilot that helps you prep for interviews with specific prompts and training, and even gives you live advice mid-interview.
4. GenFM by ElevenLabs is basically ElevenLabs’ competitor to NotebookLM (lets you generate a podcast for your docs).
5. BoostSpace creates a single source of truth for your data and connects this dataset with all your other tools (great video—very campy).
6. Perplexity Shopping is a standalone shopping app that’s like a search engine for products (pro feature but its free to try for non-pro users).

See our top 51 AI Tools for Business here!

*This is sponsored content. Advertise in The Neuron here.

Under the Hood.

Really cool project - here are the details. Or, if you prefer, you can try it directly in their INTELLECT-1 chatbot here.

1. QwQ-Preview is Alibaba’s answer to OpenAI’s o1 reasoning model—you can run it on HuggingFace, and use it for commercial purposes under Apache 2.0, but it’s definitely a preview as it does some weird stuff like change languages.
   1. Fun fact: OpenAI is now trying to trademark OpenAI o1 (paperwork here) to shield its IP from copycats.
2. AgentAuth helps you connect your AI agents with other apps.
3. Someone made an open-source version of OpenAI’s canvas (code here).
4. llama.cpp is an inference engine to run large language models locally on your own hardware (guide here), though many users prefer more user-friendly tools built on top of it like Ollama or LM Studio for simpler deployment.

Sunday Special

AI Will Cause Jobs to Evolve, Will You Be Able to Evolve With It?│Justin Wolfers

Above is the economist Justin Wolfer’s take on what job skills will be valuable in an AI future. The TL;DR is physical labor is the safest (but it doesn’t pay well enough atm), so “science and critical thinking skills” is your second best bet.

The skills to avoid investing in? Writing and coding…

BRB, looking up cheapest + fastest science degrees to get for noobs…

Watch it for yourself, though—the whole video is great!

A Cat's Commentary.