Interesting Read!

I did something similar with f5bot but this one is more structured and cleaner.

In this example, we’ll build a basic Agentic RAG system using:

VectorStore: Retrieves relevant information from a pre-indexed document database. WebSearch: Fetches up-to-date data from the web when VectorStore lacks the required information. The AI agent dynamically selects the appropriate tool based on the query, showcasing the adaptability and efficiency of agentic RAG.

LLM: gemini-2.0-flash-exp Embedding Model: BAAI/bge-small-en-v1.5

Try it out!

Rewrite-Retrieve-Read is a three-step framework for tasks that involve retrieval augmentation, such as open-domain question answering. It focuses on improving the quality of retrieved information and generating accurate outputs by refining the input query.

The fact that ppl think AI is now the solution to everything just because we have a good auto complete tech developed goes to show how much our industry needs a reset. The other day I read someones comment about how we should spend 7T dollars to get to AGI. If our industry had a little more sense and standard, half of the ppl in tech wouldnt meet the bar for sensibility. We probably would also be a lot further ahead in areas we need to be and id definitely enjoy working more in tech. Less idiocy, more sensibility

you don't really need "AI". There are plenty of solid tools already available for archeology, typically what is lacking is the images to run them on. That's why they are still discovering entire "lost cities" - because someone takes good quality pictures for one reason or another, archeologists then feed them into their tools, and voilà...

Good Post. But I struggled to understand it a first. Would be really helpful if you add a short description explaining it.

It's simple: it starts with Scheme being implemented in Prolog and then Kanren gets implemented in the Prolog Scheme. It's much less difficult than a couple of magnets and an air hockey table.

Very Sad to see Claude Sonnet not included. Deepseek is now a good competitor too. Hoping to see something similar for both of them. Is there anything available for it already?

Good Game. Was it created using some AI Model?

I used AI, but mostly for geometry (e.g. find a formula to understand that lines are crossing each other).

Close, we're training an AI model, profiting a mTurk and bypassing captchas

This is an awesome resource. This guy Shubham posts great stuff on twitter and linkedin too. Check out in case you haven't.

I have got a lot of my LLM resources through his page. He is a more of a LLM curator and less of a creator but highly recommend.

This is just amazing. I tried out with my startup's repo and it was a blast. Shared with my community

thank you so much!!