How to Make Money With AI using no-code and low code

This year has seen an absolute explosion of new AI capabilities with the launch of GPT-3.5-turbo, GPT-4, Anthropic’s Claude, Google Palm, and now Llama 2. Because AI is improving so quickly and can now unlock real business value there’s a thriving market for entrepreneurs to take advantage of. Businesses and consumers are going to continually adopt AI solutions at a staggering rate as demonstrated by market studies. The market for AI solutions is expected to grow from $100 billion in 2021 to over $2 trillion by 2030. In this article we go over some ways that entrepreneurs and businesses can make money using AI. Some of these ideas require technical skills while others can be delivered using no code tools.

Build Custom Chatbots for Companies Using No-code

There is a huge demand from businesses to adopt automated customer support bots as evidenced by the success of chatbots like SiteGPT and Chatbase. One powerful and relatively easy opportunity is to reach out to businesses in your city and offer to help them with automating customer support using custom chatbots. These type of GPT powered chatbot applications can be built using drag and drop tools like Botpress or Typebot. You can bill a one time fee plus a monthly retainer to cover the cost of maintenance. Check out the below image of the botpress studio to build chatbots. You can also find several youtube tutorials on how to get started with botpress.

Build LLM powered apps using Typeblock

Maybe I am biased here, but this is probably the easiest way to make money with AI. Typeblock is a no-code tools that lets you build and share GPT powered applications using a Notion like WYSIWYG. If you have dreams of spinning up an AI powered SaaS you can do so with Typeblock which handles all the user authentication, payments, and custom domain setup for you. If you need ideas just think of this one example: a company called Venturusai.com that simply evaluates business ideas using GPT-3.5 and GPT-4. This idea can very easily be replicated using Typeblock. Venturusai is just a few months old and is doing more than $2000/month. They also recently received an angel investment from Jason Calcanais. If you want to build GPT-3.5 and GPT-4 powered solutions that can access audio, the internet, PDFs, CSVs, and create value for users Typeblock is a great option. Here’s a video on how to set up payments in Typeblock in under 4 minutes. Also check out this voice to blog app I made in 2 minutes.

Start an AI tools directory

No doubt there are a ton of aI tools popping up everyday. According to Replit they saw a 34000% increase in new AI projects built on their platform. Starting an AI tools directory could be a great business opportunity new AI tools are being created everyday and the growth is clearly exponential. Although there are currently several popular directories like futurepedia, there still is a need for better curation, categorization, and search. A huge plus could also be user reviews. You could make money by asking companies to pay a small fee to be featured at the top of the directory listing for a few days. To create something like this using no-code I would use either Webflow or Softr. Here is a softr template you can check out: https://betadirectories.softr.app/

With this space I would be especially interested in building a tool directory for chatGPT plugins as the current experience for finding plugins is quite dreadful. Check out plugin.surf for a good example. I believe there is definitely space for more players to help user find and evaluate useful plugins.

Help companies fine tune their own models

Offering AI model fine-tuning as a service is a potentially lucrative idea because many companies are investing in AI but struggle with model optimization, additionally they want their models to be trained on internal company data. Fine tuning models allows you to train a model on a dataset so that it becomes better at solving a specific task.

Fine tuning use to require extensive AI expertise but now it has become a lot easier thanks to companies like Replicate and OpenAI. You will still need to know some very basic coding skills, but it can all be learned by watching youtube tutorials and reading online documentation. For example check out the step by step instructions on replicate’s website on fine tuning. With replicate you can fine tune both text based and image based models. The text based models that are currently supported are

1. flan-t5-xl
2. llama-7b
3. llama7b-v2-chat
4. gpt-j-6b
5. llama-13b-lora
6. stablelm-base-alpha-7b
7. stablelm-base-alpha-3b

You can also fine tune Open AI models by following the instructions on Open AI’s site. Note that GPT-3.5 and GPT-4 are not currently available for fine tuning, but should become available at the end of the year if you want to stick to helping companies fine tune models from OpenAI.

Create AI girlfriends based on influencers

This one is probably the most controversial, but potentially the fastest way to make money. Just check out the google trends search volume. Just check out Caryn AI which made $70,000 in just one week. What I would do is go to influencers and partner with them to build them digital AI replicas trained on their content and using their voice. If they have a youtube presence you can easily get the youtube transcripts and embed the transcripts using a vector DB. Note that you will need to moderately technical to be able to execute this idea, however, there are so many youtube videos on the underlying tech that anyone dedicated could learn. You can then use a voice cloning tool like ElevenLabs to create a realistic sounding replica of their voice based on the audio from the videos. You could take a cut of the revenue by partnering with them in exchange for handling all the tech setup.

Datasets as a service

Offering customized data collection and dataset creation for AI model training could be a good opportunity, because every company that wants to build or fine tune their own models is going to need data. And a lot of it.

There’s at least two ways to get data. One is through scraping and the other is using another AI model to create a synthetic data set. For data scraping you could write your own scraping scripts, which is actually not that difficult. It’s also made a lot easier because you can ask chatGPT to help you write these scraping scripts. Check out this snippet for example. You could also use a no-code scraping tool like Browserbear. The other option is to actually create these datasets using chatGPT. Here is an example from chatGPT on how you would do so. Interestingly this is how the LLM Alpaca was trained.

You can also obtain datasets from online resources like Kaggle. If I were interested in this idea I would consider first starting an agency and approaching companies to see if they would be interested in custom models trained on custom datasets. Over time you could probably reuse or resell some of these crafted datasets.

Help content creators repurpose their youtube videos/podcasts

Content creators spend a ton of time recording and editing. So much so that after the episode has been posted they are exhausted and don’t have the energy to create marketing collateral for other social media channels.

One option is to create a tool that takes the youtube transcript and then coverts that content into the collateral social media content. You can actually easily do this with a tool like Typeblock, email me at toshvelaga@gmail.com if you need help setting this up.

You could also use Whisper from OpenAI if it’s a podcast and upload the content to Typeblock to turn it into a LinkedIn post. You could also go one step further and automate this by hooking into Zapier or Make that that automatically posts the content to social channels once you’ve posted to youtube.

Helping content creators repurpose their video and audio content into optimized written formats like blog posts, tweets and LinkedIn updates using AI technology represents a promising business opportunity. AI automation can handle the most manually intensive and time-consuming parts of the process like transcription and drafting initial versions. This significantly reduces turnaround time compared to purely manual repurposing, allowing the business to scale and serve more clients faster.

Summary

Hopefully you found these ideas helpful. By no means is this list exhaustive, but this should help get you going to think about the ways in which you can make your mark as an entrepreneur in the exponentially growing market of AI.

To recap the ideas we covered were custom chatbots, custom LLM powered tools, AI tool directories, fine tuning models as a service, AI girlfriends, collecting curated datasets for models, and creating collateral marketing content for youtubers. All these ideas can be done using no-code tools or some light coding. For the light coding tasks most of these skills can easily be learned and applied using chatGPT and youtube tutorials.

If you see an idea that particularly speaks to you or you think I am missing something feel free to reach out to me at toshvelaga@gmail.com. Happy to talk more about any of these ideas and how to implement them.