6
53 Comments

Blog (SEO) on autopilot. Roast my idea ๐Ÿ”ฅ

As a developer what I like most is to create new products, what I hate most is to do marketing about my products.

For one of my last projects I created a Blog to improve SEO on autopilot.

How?

I created this python script that run on a server and do scraping about news and specific rss feed. When there is something new, it sends the data as a context to OpenAI API and create a new blog post written to improve the SEO of my site.

Right now Iโ€™m testing it. I will share the results in a few weeks.

IDEA ๐Ÿ’ก

What Iโ€™m thinking to do is to create a service on top of it.

It would works something like that:

  1. Create an account
  2. Choose the streams of source (rss feed, google alert, etc)
  3. Add some customization (cta, description of your website, link)
  4. I will create a custom blog for you and every time there is a new content from the data source it will publish a new post automatically
  5. Using the DNS you can link your website domain to the blog page

Note: of course you can approve the articles before it is published and change it with a WYSIWYG editor.

Note 2: the article will also have a custom cover, all the meta tags and a sitemap of the articles.

Demo (sorry it is in Italian only ๐Ÿ˜ข): https://bandofacile.com/blog

Feedbacks?

What do you think about it? Is something you may be interested in?

Roast this idea ๐Ÿ”ฅ๐Ÿ™๐Ÿป

on July 30, 2023
  1. 2

    This is something mandatory for the current SEO world.
    How about the accuracy of the response?
    GPT4 would be right option for commercial purpose and in that cases, how will the pricing determined ?

    1. 1

      I m using the API of OpenAI with a temperature of 0, so it means that it just elaborate the info you give.

      Btw Iโ€™m building an MVP ๐Ÿ˜ would you pay $29/month for this service?

    1. 1

      Thanks Alanso! Iโ€™m building an MVP ๐Ÿ˜ would you pay $29/month for this service?

  2. 2

    I made something like article generator but for podcast. So people can generate the article from the podcast episode. The article outcome is quite decent ( sample articles : https://medium.com/@muhdazharali98 , and do get bunch of claps on daily basis lol )

    The trick that I found is don't let the ai write the article end to end provide information ( since you already scrapping internet ) and ask it to write into specific topic. Else it will just scratch the surface and sound so much like AI with useless Article. So try make it so it focus on specific topic.

    For example if you are talking about what place to visit in Thailand, do not ask the AI to write about "place to visit" give it one specific place and give it the why and how to visit the place and then ask the AI to make article. So the title will become "why you must visit Phuket when going to Thailand" and you provide the information

    You can out my tool it is free and let me know what you think : decentool.com .

    1. 1

      This tool decentool.com doesn't give accurate result ... Video was something else and it gave completely different article !

      1. 1

        if you dont mind can I know the topic and video that you used

    2. 1

      Like it! Really interesting ๐Ÿ˜

  3. 1

    I will definitely try this approach, it is a win win situation to write content easily on the latest trends in your preferred niche.

    Thanks for the idea Riccardo. Would

    1. 1

      Hey @zipyteam the MVP is almost ready.

      I'm not looking for general feedback, but more of a "would you pay $29 per month for this service?"

      1. 1

        Not right now, will get in touch if something similar comes up. Thanks!

    2. 1

      Thanks for the feedback ๐Ÿ˜ actually I am working on an MVPโ€ฆ are you interested in testing it as soon as it is ready?

  4. 1

    Love this idea.

    Definitely want to have such a tool at my disposal :)

    As someone who has been trying to stay on the publishing treadmill for the past 18 months, would jump at anything that helps me stay the course, effortlessly and efficiently so.

    But at the same time have been disappointed many times with mediocre outputs :(

    I feel AI isn't still there. It's definitely close, but still not there. It could be any day!! so, I keep experimenting every once in a while.

    That said, with the idea, the possibilities are endless - could expand into tweets, newsletters, etc etc.. but need to find that RIGHT MIX first.

    1. 1

      Thanks for the feedback! I think AI will get better and better, I want to jump on this train now and be ready in the future.

      I am a (lazy) dev, so automation is my focus number one ๐Ÿ˜‚

      Are you interested in testing this as soon as it is ready?

      1. 1

        Sure. Count me in :)

        Even now, as we chat, I'm struggling to have chatGPT produce the output to my liking.

        1. 1

          The MVP is almost readyโ€ฆ working on some deeper validation before launching!

          Question:
          Would you pay $29/month for this service?

          1. 1

            YES, if (1) it gives me the kind of output I'm looking for ans (2) at equal or lesser cost than I'm incurring now.

        2. 1

          I read somewhere (canโ€™t find the link right now) that OpenAi is creating a free prompt engineering courseโ€ฆ I would jump into it immediately

          1. 1

            I had started one in April. This one - https://learnprompting.org

            I guess I need a refresher :)

            While I get good, as-desired short form outputs, long form is always a struggle

            1. 1

              In my experience it's better to split the task in shorter ones.. the results are much more better... and if you can use the API instead of ChatGPT, it's even better

              1. 1

                agree ๐Ÿ’ฏ - about the shorter tasks.

                I have no experience using the API. Will try it out soon. Thanks

  5. 1

    I've thought about this too, even written a few articles myself with GPT 4. I find that they're somewhat lacking and not very useful. Maybe my prompts/context just suck though.

    Whenever I search google for articles that I actually read, they're usually a very detailed comparison and talk about experiences ie. "why I chose to use render over fly.io" or something like that. If you can replicate articles with those experiences then you've got a winner.

    Also keep in mind that just pumping out junk articles won't be good for SEO, make sure they actually get clicks and prune the ones that don't.

    1. 1

      I agree with you. I am improving the prompt dailyโ€ฆ i will share the results in few weeks ๐Ÿ˜

      (An MVP will be ready in daysโ€ฆ but for the SEO results it takes more time)

      1. 1

        The MVP is almost readyโ€ฆ question: would you pay $29/month for this service?

    1. 1

      @elliottzelinskas the MVP is almost ready ๐Ÿ˜ would you pay $29/month for this service?

    2. 1

      Thanks. Are you interested in testing this when ready?

  6. 1

    Idea to optimize on SEO is good. Maybe that are two things. Generating a text and optimizing for SEO. It might be easier to start with already written text and then optimize a pipeline to transform it to a really good SEO driven text without losing the meaning of the original text

    1. 1

      Yeah. This is the reason why I want to start from an external source (google alert, rss feed etc)

      1. 1

        That sounds great!

        1. 1

          @slyracoon23 I am almost ready to launch the MVPโ€ฆ question: would you pay $29/month for this service?

  7. 1

    This is a needed product ... Although there are some similar services I guess on WordPress too as plugins and Honestly posting the update isn't enough for marketing !

    1. 1

      I agree. What would you like to have more than that?

      1. 1

        Engagement to that post is also imp

        1. 1

          Like comments or what?

  8. 1

    This is actually very common these days. If done carefully it can definitely work.

    1. 1

      Thanks. Are you interested in testing the MVP when ready?

      1. 1

        TBH, I have my tool that does the same ๐Ÿ˜… Not interested in releasing it tho

        1. 1

          Is it working? How are the results?

          1. 1

            It crashed ๐Ÿ˜…

    1. 1

      Iโ€™m building an MVP ๐Ÿ˜ would you pay $29/month for this service?

  9. 1

    Blog articles generated by AI are currently terrible and negatively affect brand perception (if they work in generating traffic, which is not even given). I wouldn't recommend the direct. On the other hand, making it easier to find a topic and start the article may help busy entepreneurs just finish/personalize the draft.

    1. 1

      What about writing content based on some context? They are not so bad imo

      1. 1

        I don't know. I think AI can be a useful part of the workflow, but I don't believe in using it end-to-end to produce content in a way a founder can be proud of (content farms for affiliate marketing may be a different story). Show me good articles that were 100% AI-written and I'll reconsider.

  10. 1

    I wouldn't trust AI in this case. How will it improve my SEO? What about my visitors who will see my "poorly edited weirdly worded random" writings (not all of them will be terrible, but the risk is there)? What if somehow AI generates fake news and I post it on my site?

    1. 1

      I see your point. I am testing it on my own site and as regards the hallucinations, they are not there if I choose a temperature of 0 on the openai api, which means that the AI just use the info given, without writing anything else.

      Moreover it create just a draft of the content. I receive a notification when something new is written and I can approve or edit it.

      I am scaling the google search page position thanks to those articlesโ€ฆ but I am working on just since last week, so I will need some more time for the results.

      1. 1

        It needs some time and testing for sure.

Trending on Indie Hackers
Iย spent $0 on marketingย and got 1,200ย website visitors -ย Here's my exact playbook User Avatar 41 comments Why Early-Stage Founders Should Consider Skipping Prior Art Searches for Their Patent Applications User Avatar 22 comments I built eSIMKitStore โ€” helping travelers stay online with instant QR-based eSIMs ๐ŸŒ User Avatar 20 comments Codenhack Beta โ€” Full Access + Referral User Avatar 20 comments Veo 3.1 vs Sora 2: AI Video Generation in 2025 ๐ŸŽฌ๐Ÿค– User Avatar 18 comments Day 6 - Slow days as a solo founder User Avatar 13 comments