i will go with the first solution, which if I understood correctly, involves me setting up a python project/tool, writing articles in my preferred markdown/latex, converting them and submitting them to medium right ? Instead of a one of solution, I have decided to be creative. I want to have my own pipeline. I want to have the following : 1. The ability to write text, without need to enforce any latex/markdown conventions. 2. Have my own “intelligent” word/spelling or syntax ai assistance… for example, I would want to know if my paragraph makes sense or could be improved. 3. The interface between myself and the system should be a metacognitive orcherstrating llm assistant…I do not want any of the assistants incessantly suggesting anything or reading/correcting anything, they should only do so when invoked. Otherwise, if there is a problem, the metacognitive orchestrator will solve it before compilation. 4. After the article matches the requirements of the meta orchestrator ( there should be a diagram here, a formula there etc), it is then published automatically on medium. Each article should start as a task, whereby the user describes as much as they can, what they expert the article to be like. The meta orchestrator then has a list of items to ensure are present.

Are there no other solutions that offer all the features I mentioned, with better functionality and better codebase ? No, because, when I am writing an article, I ALWAYS, ALWAYS, ALWAYS need to be combining material that could be available, but is not there. Say for example, I am writing about the Pugachev Cobra Maneurvour: this requires me to make regular google searches (to get to wikipedia, I am too lazy to have it pinned, its always the first result) and one on google scholar. That is two internet tabs I have opened, and am still not anywhere closer to my answer. I then have to read the page looking for an information node, and what would be the best node here ? Pugachev right ? wrong, that name is not mentioned anywhere on the Wikipedia page. It is only on an article on Research Gate will one find the exact terms “Pugachev Cobra”. This is too much work for me to use that reference in an article, and I will not, because its too much work. What if, I could have a simple “programming language”/programming interface, that works sort of the following :

If microsoft word exactly like excel, it wouldn’t be the arm pit that it is, even though excel is bum sweat. What if we took writing and made it sort of like excel, that is, sentences can form “cells/units” that can be operated on. For example, if I selected a text paragraph and then added some few commands in this language, the text could then be operated on, as a unit, like a number. Finding if the grammer is correct would be akin to finding if a number is a say, a prime. One could then do more sophisticated procedures like for example, partition it, make online searches and store the knowledge from the searches in a “knowledge database”..what does that mean ? looking back at our Pugachev Cobra example, the information gathering should be automated, and stored as a “knowledge source”…so that, for example, one could query that source for specific themes like “zoom climb, supersonic travel, maneuvrebility”…the metacognitive orchestrator then allows the user to see results/information from the search knowledge base that is useful in attaining the goal of the said text. That is the most crucial part. If the essay was on high altitude respiration, and a user searched for “Pugachev Cobra”, the results they should get should only be about pilot respiration whilst performing a Pugachev cobra.

This package is not meant to be commercial tool or to meet some other standars/criteria other than the ones described above because they are not provided anywhere, if they are provided, in their full form, and do not cost money, I am ready to use them. I expert to use free specialised models which one can easily access via huggingface api.