adaboese

I appreciate this conversation. It gives me ideas how to think about the future copy of the landing page. Truth be said, that landing page and most of the blog posts were made haphazardly in a single afternoon. Heck, many people think that AIMD is only used for generating copy for comics... It is well overdue for a refresh based on all the learnings.

Just migrated to Google Cloud last night. Still ironing out some details. Will check what broke there.

RE terrible sources, the sources are chosen based on what Google identifies to be credible sources. I am aware that some of the sources Google chooses as the source of truth are not great though. I've not found a good way to filter them out.

Out of curiosity. Which source in particular stood out as low-quality?

Well, I am not here to win Internet arguments. Try the product to form an opinion, if you feel strongly to discredit what I am saying.

As mentioned before, sales gen (which is the blog post you shared) and product are not going to match 1:1. You have to reach the audience at the points that you know they research and try to solve. That's not going to be a surprise to anyone who built and scaled products.

How does that model rank the title I shared?

The secret sauce behind "anti-AI detection" is 'not writing overly generic, unreferenced content'. Not a single line in the code base tries to evade AI detection; it is just the side-effect of how the product works.

At the end of the day, my job is to sell to the ICP that I am targeting, and I know that their concern (based on user interviews) is that content will be detected as AI detected. I need to get them into the product first before I can educate them.

If that's the criteria, then fair, but AIMD produces articles that are virtually indistinguishable from what an average copywriter would compose if you hired them.

AIMD is very different from your typical article generators (like https://writesonic.com/, which good God I don't know why anyone would use, but they are doing good, so good for them) that is just spinning out LLM-written content. AIMD orchestrates a multi-step process of topic discovery, source research, content summary, fact validation, hypothesis generation, outline planning, and only then writing. Content generation is a tiny surface area of AIMD, for which I am the least excited. It is really meant to generate a data-rich starting point for writing a high-quality article, rather than produce something that's ready to post. That is also reflected in the product: the content score is based (among other variables) on the % of insights added by the editor VS generated.

You are free to judge my business model (which, if you are referring to content generation, YC is heavily invested in), but nothing about the content I've shared is spam.

Also, if you dig deeper into what AIMD does, AIMD was built to combat spam. The one-pager is a good starting point.

In what way on Earth is the post that I am referring to an SEO junk?

so re-submitting after a couple of hours is really the way to go?

Wrong link, I think. Takes to someone's FB profile

given the posting history of this account, I cannot tell if this is posted by the maintainer or the guy who created the PR.

Who is running the show?

I meant, make it open-source and allow to self host it. Have encrypted database for hosted solution.

Reach out to SaaS companies that could benefit from it.

I am constantly on the look out for models/data sources that would give AIMD an advantage. I am sure others are too.

Someone should build an open-source version of Carta. I've been thinking about it for a while.

That's really neat. I always wanted to get one, but have hard time substantiating it given that I already have Nintendo. can anyone talk from experience if it was worth it?

Is that an actual minimalist website or is this website just actually 20 years old?

so I still need to download and host models myself.

I found that to be incredibly impractical. I tried to do it for my project AIMD, but the cost and quality just made absolutely no sense even with the top models.

What is nitter?

I cannot really wrap my head around what is this. Is this the whole model? Is this some sort of middleware between user and OpenAI?

Why is there a massive @facebook tag on the page?

_oh_

I own linksoap.com, yes. Just redirecting it to Google until I start using it.

That's good feedback. Funny that a few people already mentioned LinkClean and LinkStitch.

I am trying to not to lean on "clean" too much, as the the keyword is actually interlinking (or "internal contextual cross-linking" to be precise). But it is helluva long string of words to use in a name.

Started as a stream of thought. Ended up turning it into a blog post.

This is ironic because it means that they are either using AI to generate their blog (fine, but ironic) or that their AI detection too does not work.

I am building an AI content generator and my edge that makes it better is sourcing data from not easily accessible data sources (think research papers) through SEO. I cannot imagine that these companies are not using every bit of confidential data for their advantage. Proprietary data access is the only thing that matters in the future of AI wars

Misread as "AI writer". I thought it is another competitor

Related to this, does anyone a good way to search only for articles?

I have a use case that requires that I only surface search results that are articles (news, blog posts, etc) rather than anything else.

The best I've gotten to so far is doing queries such as inurl:blog

I would never notice this. Seems such an edge case that it would only pollute the namespace of properties. There are already so many properties to learn, and something that you will use once in your lifetime does not belong in the language itself.