Since ChatGPT was released in 2022, there’s been an explosion of AI chatbots and assistants. There have been just as many, if not more, tools using LLMs being developed.
Some of these are just tools for running LLMs locally, like LM Studio.
Others are power tools that make chatbots a lot more useful by giving them access to your files or databases, giving them tools to extend their functionality, or allowing access to your system, like Newelle or Anything LLM.
Then there are tools that create entire AI workflows and agents powered by LLMs, such as Microsoft AutoGen, Dyad, and ChatDev.
These tools are cool, innovative, and useful, but there are way too many of them to figure out which ones actually work and are worth the time, and which, while cool in concept, don’t actually work, or are too complicated to use.
I’ll be posting about the ones I’ve tried out, along with ones that I haven’t gotten to yet but sound promising. If you have any experience with these or others, please let us know!
LM Studio is a desktop app for running LLMs locally.
I’ve run it on Windows, and it was pretty cool to have LLMs running 100% offline.
The setup is very easy, just install the .exe, and you can then download models from within the app, or import GGUFs using its easy to use CLI/TLI, and you’re good to go.
I’ve run Deepseek r1, Gemma, and Llama in it.
LM Studio is very user-friendly and has some great features. It supports:
Speech input, allowing you to talk to your AI.
Vision models, which let you ask questions about images you upload.
Uploading files, so you can chat with your own documents and PDFs.
The app’s focus is on providing a clean, ChatGPT-like interface for chatting with AI. While it’s not designed for other AI tasks like image generation, it’s a perfect starter for
anyone wanting to experiment with local LLMs.
Point is, LM Studio is an easy-to-use, simple-to-setup, and reliable application for running LLMs locally, making it a great choice for anyone looking to get started with offline AI.
Microsoft AutoGen is a tool that lets you build AI apps with multiple AI “agents.” These agents can talk and work together to get a job done, like a human team.
Autogroq is a tool that makes AutoGen easier to use. Instead of you having to set up the agent team yourself, Autogroq automatically creates a specialized team of AutoGen agents just from your prompt.
Using Autogroq is simple: you provide a prompt describing your task, and Autogroq handles the rest. It uses your prompt to figure out what kind of experts are needed, creates the agents, and then sets up a workflow for them to collaborate and solve the problem. You can then watch the agents work and talk to them to guide the process.
At least, that’s how it’s supposed to work.
In reality, Autogroq seems to be broken and incompatible with modern versions of AutoGen right now.
I spent hours debugging it, trying to figure out why workflows weren’t getting exported to AutoGen, and why it was giving me ZIPs when AutoGen only takes JSONs, until I realized that Autogroq isn’t compatible with newer versions of AutoGen.
In summary: Don’t use it, unless its dev updates it.
Dyad is a free tool that allows you to build apps using AI. You just tell it what you want it to build, and then let AI do the work.
It builds web apps, using Node.js.
Installation and setup is pretty easy.
You just download the .deb or .exe from the Github releases page, and install.
You’ll have to manually install node.js, give it an AI API key, and then you’re ready to use Dyad.
You can do everything right within the app, including:
chat with the AI
have it build your app
run it
debug it
preview it
Dyad is an impressive, all-in-one tool for quickly building and testing web apps with the help of AI.
It’s an innovative, powerful, and revolutionary tool, and I recommend you go check it out right now.
Everywhere is an interactive AI assistant with context-aware capabilities, featuring a sleek, modern UI and powerful integrated functionality.
Unlike traditional AI tools, Everywhere instantly perceives and understands anything on your screen.
No need for screenshots, copying, or switching apps—just press a shortcut key to get the help you need right where you are, delivering seamless AI assistant support.
Usage:
❔ Troubleshooting Expert
You encounter an error message while using your computer but are unsure how to resolve it
Invoke Everywhere next to the error message
Command: “What is this error? How to solve it?”
Result: Everywhere captures the message context and provides a solution
📰 Quick Web Summarization
You’re browsing a lengthy technical article but need just the key points
Invoke Everywhere over the webpage
Command: “Give me a 3-point summary of this”
Result: Instantly receive the main arguments without reading the entire content
🌐 Instant Translation
You encounter text in a foreign language while researching
Invoke Everywhere over the text
Command: “Translate this to English”
Result: See an immediate translation without switching to another app
✉️ Email Draft Assistance
You’re writing an important business email but unsure about the tone
Invoke Everywhere over your draft
Command: “Make this sound more professional”
Result: Your casual text is transformed into polished business communication
Usage:
Invoke Everywhere next to the error message
Command: “What is this error? How to solve it?”
Result: Everywhere captures the message context and provides a solution
OpenAI
Anthropic (Claude)
Google (Gemini)
DeepSeek
Moonshot (Kimi)
OpenRouter
SiliconCloud
Ollama