Rami's Readings #84 - 🆕 List of LLM Agent Libraries

The latest on the AI, LLMs, RAG, WordPress, Dubai, and more.

Welcome to Rami’s Readings #84 - a weekly digest of interesting articles, papers, videos, and X threads from my various sources across the Internet. Expect a list of reads covering AI, technology, business, culture, fashion, travel, and more. Learn about what I do at ramisayar.com/about.

---

---

🤖 AI Reads

In this edition of the newsletter, I’m trying something new: a curated list of LLM, agent, and workflow orchestration libraries and platforms, accompanied by brief notes.

List of Generative AI (LLM) Orchestration & Agent Libraries / Toolkits / Services

Mostly focusing on libraries for building LLM-powered apps / systems. Sorted in no particular order. OSS: Open Source Software. This is not a sponsored post.

1. LangChain (OSS & Managed Service)

   Notes: Provides native Python abstractions. Supports production-ready workflows. Strong third-party integrations. Offers a managed deployment service currently in beta. VC-backed.

2. Mirascope (OSS)

   Notes: Provides native Python abstractions. Weak support for Agent-based and Graph/Workflow-based AI. Strong third-party integrations.

3. Haystack (OSS)

   Notes: Provides native Python abstractions. Supports production-ready workflows, RAG, etc. Strong third-party integrations. VC-backed.

4. LlamaIndex (OSS)

   Notes: Provides a complete platform from data ingestion to observability. Strong third-party integrations. VC-backed.

5. PromptFlow (OSS)

   Notes: Provides a complete platform from data to monitoring. Supported by Microsoft.

6. Rasa (OSS & Managed Service)

   Notes: Provides native Python abstractions. Supports production-ready workflows, RAG, etc. Strong third-party integrations. Provides a no-code UI and offers a managed deployment service.

7. AutoGPT (OSS & Managed Service)

   Notes: Provides a platform for creating low-code agents, ML workflows, monitoring, etc.

8. Guidance (OSS)

   Notes: Provides a native Python programming paradigm for controlling LLM systems – can be used to build AI workflows for any application.

9. MetaGPT (OSS)

   Notes: Provides a multi-agent framework with natural language programming.

10. Instructor (OSS)

    Notes: Provides native Python abstraction for structuring LLM outputs.

11. Priompt (OSS)

    Notes: Provides prompt-engineering toolkit using JSX.

12. Swarm (OSS)

    Notes: Toolkit for multi-agent orchestration from OpenAI.

If you found this useful, please fill in this poll. Your feedback is appreciated!

Pixtral 12B

Notes: Mistral AI released a paper about their Pixtral 12B model. The model was released a few weeks ago.

Astute RAG: Overcoming Imperfect Retrieval Augmentation and Knowledge Conflicts for Large Language Models

Notes: Interesting paper from Google Cloud.

Long-Context LLMs Meet RAG: Overcoming Challenges for Long Inputs in RAG

Notes: More from Google Cloud.

ToolGen: Unified Tool Retrieval and Calling via Generation

Notes: Neat.

Artificial intelligence-powered chatbots in search engines: a cross-sectional study on the quality and risks of drug information for patients

Notes: Great paper shared by a friend.

💼 Business Reads

WordPress.org’s Latest Move Involves Taking Control of a WP Engine Plugin

Notes: This move is insane - akin to Apple taking over an app on the App Store. If I ran a WordPress company, I would start ramping up my team on alternatives (like Gatsby). Read my previous comments.

Dubai’s Allure to Expats Is Weighing on City’s Infrastructure

Notes: Yes, this was confirmed by my colleagues and friends.

Photo by Christoph Schulz on Unsplash

Signing off from Redmond.