mark_l_watson (@mark_l_watson) 's Twitter Profile
mark_l_watson

@mark_l_watson

AI Generalist. Author 20 books on AI, LLMs, DL, semantic web, Lisp. 55 patents. My recent books are free to read or buy online leanpub.com/u/markwatson

ID: 16856664

linkhttps://markwatson.com calendar_today19-10-2008 21:45:12

2,2K Tweet

2,2K Followers

752 Following

mark_l_watson (@mark_l_watson) 's Twitter Profile Photo

I just wrote a short book that I hope people find useful titled "LM Studio In Action: Building Safe, Private AI with LLMs, Function Calling and Agents with MCP" as usual you can read it free online leanpub.com/LMstudio/read

mark_l_watson (@mark_l_watson) 's Twitter Profile Photo

The Meilisearch local search engine platform (written in Rust) is fantastic. I love running local LLMs using Ollama and LM Studio and the missing part of the puzzle for me is the requirement to use third party search engines. Most of the local LLM/search interactions I

mark_l_watson (@mark_l_watson) 's Twitter Profile Photo

There is a tiny change in my Clojure AI book: For my own experiments and research I have started using Google's Java SDK for Gemini rather than my old REST library. I added a section to the end of chapter "Using the Google Gemini 2.5 Flash APIs" with the new tiny library that

mark_l_watson (@mark_l_watson) 's Twitter Profile Photo

This is also my thought exactly: perceived problems with GPT-5 result from engineering decisions to make hosting GPT-5 more cost effective. TBH, I think ‘anti-dumping‘ laws should apply to all AI vendors who sell below cost. That is one reason that I have more or less switched to

mark_l_watson (@mark_l_watson) 's Twitter Profile Photo

Two changes to my Ollama book: 1. I now use a single GitHub repo for both code examples and the manuscript files: github.com/mark-watson/Ol… 2. New chapter: "Building Agents with Ollama and the Agno Agent Library"

mark_l_watson (@mark_l_watson) 's Twitter Profile Photo

I am a fan and long term supporter of ProtonMail. I have been playing around with their luma private and secure LLM chat app. I asked it what models it uses and it said Mixtral 8x7B and Mixtral 8x22B and it clearly explained how its integrated search works. The integrated search

mark_l_watson (@mark_l_watson) 's Twitter Profile Photo

Some tech posts on Mastodon often take me down memory lane, for example a post I just saw on the Fediverse. I have been so far down the AI / LLM rabbit hole the last three years that the free and open networking part of my brain had gone to sleep but I just used Lumo

Maria Popova (@themarginalian) 's Twitter Profile Photo

"Solitude is not chosen, any more than destiny is chosen. Solitude comes to us if we have within us the magic stone that attracts destiny." Beautiful read from Hesse themarginalian.org/2019/01/15/her…

mark_l_watson (@mark_l_watson) 's Twitter Profile Photo

I updated my book "A Lisp Programmer Living in Python-Land: The Hy Programming Language". Read online: leanpub.com/hy-lisp-python… This version of this book (updated in August 2025) has major code changes: - Code examples modified to work with the latest Hy version 1.1.0. - The

mark_l_watson (@mark_l_watson) 's Twitter Profile Photo

Fantastic synopsis of Oliver Sacks on perception: our own consciousness and identify dictate how we percieve the world and what we effectly choose to attend to:

mark_l_watson (@mark_l_watson) 's Twitter Profile Photo

After writing ten books for traditional publishers (J Wiley, McGraw Hill, Springer Verlag, etc.) I switched over to self publishing books using one of the Creative Commons licenses. I was just reviewing the specific license I use, and did a nostalgia deep dive: I was the Featured

mark_l_watson (@mark_l_watson) 's Twitter Profile Photo

I really like Gemini's new built-in tool "url_context" - fits several use cases and makes things easy to set the content of one or more web pages as context for a prompt. Here is a short example I added for the Hy (hylang) language: leanpub.com/hy-lisp-python…