Author: Franz Malten Buemann

I pay for ChatGPT Plus. I also get Copilot with my Microsoft Business account.

Copilot is often faster than even Plus ChatGPT on stuff it can do, but a lot of the time, especially on anything touching crime or political stats, it starts responding and then hides the generated response and replies:

Sorry, it looks like I can’t chat about this. Let’s try a different topic.

The same prompts in ChatGPT don’t hit me with that, but again CGPT Plus is often slower.

What other AIs should I consider that are quick and have depth, but are less censored?

If you have an eshop with thousands of products, Ragus AI can basically take any RSS feed, transform it into structured data and upload into your target database swiftly. Works best with Voiceflow, but also integrates with Qdrant, Supabase Vectors, OpenAI vector stores and more. The process can also be automated via the platform, even allowing to rescrape the RSS every 5 minutes. They have tutorials on how to use this platform on their youtube channel (visible on their landing page)

To be clear, I am not looking for a chatbot that can do my work for me. I’m sure that’s what anyone says but genuinely I’m just looking for something that’s good at math reasoning/explanations specifically in the short term but one that’s overall not prone to mistakes as I would like to essentially find one I can use as a second search engine. Even though they’re often inaccurate in deciphering some of the info, the info aggregation they do is so convenient. That being said I suppose I’d add that being able to easily click on source links is also a plus. I’ve tested all the big ones but I can’t make a clear judgement based on what I’ve used them for.

After building custom AI agents for multiple clients, i realised that no matter how smart the LLM is you still need a clean and structured database. Just turning on the websearch isn’t enough, it will only provide shallow answers or not what was asked.. If you want the agent to output coherence and not AI slop, you need structured RAG. Which i found out Ragus AI helps me best with.

Instead of just dumping text, it actually organizes the information. This is the biggest pain point solved – works for Voiceflow, OpenAI vector stores, qdrant, supabase, and more.. If the data isn’t structured correctly, retrieval is ineffective.
Since it uses a curated knowledge base, the agent stays on track. No more random hallucinations from weird search results. I was able to hook this into my agentic workflow much faster than manual Pinecone/LangChain setups, i didnt have to manually vibecode some complex script.

Hey all,

Over Christmas and New Year I spent most of my time building this app. It started as something I made purely for myself.

I was constantly switching between tabs and accounts just to ask questions to different LLMs, and it was killing my focus. I tried a bunch of AI chat apps, but they were missing a couple of things I really wanted:

• having access to +120 premium place in one place
• the ability to chat with multiple models at the same time
• the ability to have models respond or “debate” in parallel

So I decided to build it myself.

I launched it 2 days ago and currently have exactly 0 users, so I’d really appreciate any honest feedback, good or bad.

There’s a free plan with 20 messages per day on some cheaper models (I’m an indie dev with basically no budget), plus two paid plans with higher limits.
If anyone wants to upgrade, you can use WELCOME20 for 20% off.

Thanks for reading, and feedback is very welcome.

PS: the app is https://omny.chat