Monetising artificial intelligence applications

How pricing works + 3 strategies to make money

1. Pay as you go
2. Credits
3. Monthly subscription

Firstly, how pricing works - tokens.

Tokens can be considered as segments of language utilized in natural language processing. In English text, a single token equates to approximately 0.75 words or 4 characters, which means that 100 tokens correspond to roughly 75 words.

Cost projections

Each model has a different cost. For example:
Davinci - $0.1200 / 1K tokens
Ada - $0.0016 / 1K tokens

By understanding how many monthly users you may have, how many requests they may make a month, an estimate on the tokens per request you can forecast cost

Monetisation

Pay-as-you-go

This is by far the simplest, users upload input and pay to get the output.

An example is Lensa AI, Users upload 10 to 20 selfies to the app and for $7.99 Lensa will create a pack of 50 images, with various art styles and backgrounds.

Credits

This involves users paying to get limited access based on how many credits they've bought to use your application.

An example here is Dall E 2
1 credit – 4 images
For the first month you get 50 free credits
Additional credits packs are $15 for 115 credits

Monthly subscription

Similar to SaaS model, you provide users access to your application in exchange for a monthly fee.

An example of this is midjourney, for $30 a month you can generate AI images with fast GPU time for 15 hr/month.


Similar to the NFT/Blockchain wave, artificial intelligence integrated solutions have seen a wave of interest.

For educational purposes, building an application can be a fun experience but you may want to think about monetisation sooner rather than later.

Monetizing AI: Pricing and Strategies for Profitability

@100DaysofNoCode  Build Weekend inspired to explore creating a companion chatbot with No Code that was trained on real iMessages.


Here is my process plus results 🧵

First, I tried to export my iMessage chat history, it was more difficult than I initially thought.  Then I decided to source new data by using  @TallyForms  to specify specific questions that need to be answered.

Before using friends and family members as testers I used ChatGPT to get some answers to train my model

A lot of the questions were deeply personal so I had to tell it to act like somebody for it to even answer. I chose Will Smith.


Then I head over to @rikudotai to train my chatbot. I simply follow the prompt and enter my questions and answers.

Always remember more *quality* data, better chatbot.


I then can natively test my bot in @rikudotai. It come out pretty sweet.


I then used @AdaloHQ to create an interface with my chatbot using the chat template with some database adjustments.

After that, I used @make_hq to connect Adalo and Riku

It's rough around the edges but we successfully created an AI chatbot built on our own custom data. Here's a preview of the final product.

https://www.loom.com/share/1665c42705d742bfa53de285d609387a

Observations & Predictions

1. Augmented personalities
Currently, ChatGPT has been trained to say "As a machine learning model, I do not have the capacity to have personal experiences". It is possible to make it (or any chatbot) believe it is ACTUALLY  the real person.

While this can reduce the tension between human and machine it can an illusion for the end user about "real" and "artificial"

2. Companionship
Chatbots like this has great potential to provide companionship to those that are lonely but suspect dependence will be dangerous.

3. The personality market
I predict the market will open for creative writers to provide chatbots with an artificial story/background that is bought to life by machine learning. Customers say what kind of personality they wat it to have and writers create artificial memories.

4. Responsibility
I have created an advice chatbot but who is responsible for the end user actions. Themselves? The chatbot? Me for creating the chatbot? Riku for creating the software for me to create the chatbot? I provided an initial dataset but I don't know what  it will say

5. Influencer & celebrity monetisation
Influencers and celebrities with a creative mindset or inspirational outlook on life can create, automate and monetise their personality and experiences at scale. Making a chatbot feel like a 1 to 1 conversation with the real person.

What are your  thoughts on the project, Artificial Intelligence and No Code?

Don't forget to subscribe to my YouTube channel for creative No Code tutorials: https://www.youtube.com/@mr.hackathon

You can read the unrolled version of this thread here: https://typefully.com/dominiconorton/x0ckgAw

Creating a custom chatbot

Voice cloning & artificial intelligence audio

🧵

We've seen significant leaps in capabilities in the generation of textual information by artificial intelligence technologies.

I've been attempting to explore where these technologies are currently in their development beyond the hype and here are some challenges with AI voice.

In my opinion @DescriptApp has a pretty accurate model with "Overdub". Expression is still very difficult for the technology and the cost for an unlimited vocabulary is $24/mo. Not too pricey depending on why you're using the service.

@__UBERDUCK__ has a range of voices you can use which is really cool but the quality between each voice varies drastically and no easy way to filter by quality. You can also clone your voice if you'd prefer.

AI Video Avatars

Currently AI generated voices are locked in a system separate to AI video avatar platforms such as @synthesiaIO The separation makes it hard to mix and match the best tools together.

Public perception

Maybe it's years of Siri, Alexa and whatever Google call their AI voice technology, that has led the public to be open to AI voices. I doubt this is for an extended period of time, do we really want to watch a hour series on Netflix voiced by AI? Maybe 1 day

Ethical concerns

As always we innovate then discuss the implications of our innovations.

I foresee this being the case with AI even more now than ever. Data ownership, copyright of generated images etc. Who owns the rights of my voice after I pass?

Transparacy

It might be a extremely hard task, but identifying where AI has been used and where it hasn't been used not just in video but image, text and audio will be increasingly important, or will it? Do people only care in specific cases?

I have lots of thoughts around artificial intelligence especially in relation to Black communities.

Go ahead and subscribe to my newsletter for more thoughts, ideas and tutorials.

https://www.getrevue.co/profile/dominiconorton

I'm experimenting with a podcast entirely scripted and hosted by artificial intelligence. Feel free to contribute to the topics discussed.

https://airtable.com/shrQPyLdpUZSrwu8W

An overview of voice cloning and artificial intelligence

The challenges of artificial intelligence for mass market adoption

🧵

The lack of understanding and trust among the general public. Many people are hesitant to use AI-powered products and services because they don't fully understand how they work or they're worried about the potential consequences.


The high cost and complexity of many AI-powered products and services. Many of these technologies require specialized hardware and software, which can be expensive and difficult to use for the average person.


Concerns about the potential negative effects of AI on jobs and the economy. As AI becomes more advanced, it could replace human workers in a wide range of industries, leading to job loss and economic disruption.

Ethical and moral concerns surrounding the development and use of AI. As AI becomes more advanced, it raises questions about how these technologies should be used and who should be responsible for their development and deployment.


Although in the last couple of months/weeks we've seen huge leaps in artificial intelligence technology the marketing around the industry and technology needs to improve to help people be more accepting of it.

You can read the unrolled version of this thread here: https://typefully.com/dominiconorton/x0qBH41

The challenges slowing down mass market adoption of artificial intelligence technology

An innovative business consultant and mentor used GPT-3 to support his mentee in sending professional emails.

Here's how he did it using Zapier, Open AI & Gmail

We're going to use it to translate meeting notes to summary

https://twitter.com/DannyRichman/status/1598254671591723008?s=20&t=fXHZ294ttcAyHOT1YrvryQ

@DannyRichman

You need two Gmail accounts, one to trigger Open AI and another to send the prompt for summary. This will make more sense later.

You'll also need a Zapier account and we're going to use the unofficial OpenAI integration: https://zapier.com/developer/public-invite/171592/c994c3ca67b28f871851ba5d86100f16/

You'll also need your Open AI API Key that can be found below.


In Zapier, we want to create a new Zap starting with new email in GMail as a trigger.

This will be the Gmail you created specifically for this Open AI automation.

Next you want to send the prompt to Open AI using the body of the email as the prompt.


After that you want to send the result of the Open AI generation back to the original sender that triggered the process.


Important

To execute multi-step Zaps you have a premium account and to utilise the unofficial Open AI integration with Gmail you have to have a business Gmail account.

I love how No Code and AI is empowering business users. This is what it means to democratise technology.

If you want more No Code tutorials don't forget to subscribe to my newsletter: https://www.getrevue.co/profile/dominiconorton

You can read the unrolled version of this thread here: https://typefully.com/dominiconorton/GOjn8LM

How to creatively integrate GPT-3 with Gmail

Are you new to No Code or Indie Hacking?

The space is huge and it can be overwhelming but a strategy that I'd like to advise is specialisation.

Here's a thread on different specialisations within the No Code ecosystem 🧵

Platform specialisation

Some platforms such as Bubble or Webflow are emerging as dominant players within the field. As such, more and more people are looking for developers with those specific skill sets. Deep domain knowledge in a particular platform can be fruitful.

Industry specialisation

A platform is nothing without being used for real-world use case. Can you leverage your background in a particular industry(s) to use No Code for companies in that industry.

Example: No Code for LawTech, FinTech, ClimateTech etc

Skill specialisation

There are different types of No Code Developers. Some have a strong emphasis on design, while others excel in database architecture. Finding a speciality that aligns with you can help you define a name for yourself.

For example: No Code Ops

Vertical specialisation

Aside from platform and industry No Code has different technology verticals. This includes augmented reality, gaming, native applications, artificial intelligence. Standing out in a particular vertical is a viable way to develop domain expertise!

Regional specialisation

The No Code community is Global. Can you be known as the go-to No Code person in your country or region? Where are my Arabic speaking No Code experts? Chinese focused No Code experts?

Client size specialisation

Many No Code developers are freelancing or starting agencies. Is there a specific client size you want to focus on? Startups? Scale ups? Maybe large enterprises?

A great place to start your journey is #100DaysofNoCode

In 100 Days you'll learn a general overview of the ecosystem in bite sized chunks and complete the time with a portfolio of impressive projects and collection of new friends.

I'd love for you to learn more about No Code in my free newsletter: https://www.getrevue.co/profile/dominiconorton

Retweet this thread, you never know who it would help.

You can read the unrolled version of this thread here: https://typefully.com/dominiconorton/Jpos5oQ

A thread on different specialisations within the No Code ecosystem

Some believe that No Code is all hype so here are mainstream platforms that have launched or acquired No Code & Low Code tools

A thread 🧵

Salesforce

Lightning Flow by Salesforce allows 150000+ companies develop process driven applications by connecting workflows with third party applications.

Apple

Apple's little known enterprise development subsidiary, Claris, developed Claris Connect to helps users to build new custom apps through integration with services such as Trello, Slack, DocuSign and Box, among others, without writing any code.

Amazon

Amazon empowers over 1 million active users and developed Amazon Honeycode to help their customers build custom business applications

Google

Google acquired AppSheet to democratize development for citizen developers and even leverage Google's ML & AI capabilities.

Microsoft

We all should be familiar with PowerApps as Microsoft has found a way to empower everyone the power to build the apps they need with advanced functionality and even professional developers the tools to seamlessly extend app capabilities with Azure Functions.

IBM

IBM understand better than most that businesses are dying for hyper automation so consistently develop products to meet that demand such as their Robotic Process Automation tools.

JotForm

JotForm have over 5 million users and have cleverly developed a No Code app builder to compliment their form builder services.

No Code for Enterprises is a specific vertical of the market nonetheless the most innovative companies understand the trend of abstracting away from the codebase to democratise development for business use cases.

If you found this thread useful, subscribe to my newsletter where I post No Code tutorials, news and interesting insights: https://www.getrevue.co/profile/dominiconorton

You can read the unrolled version of this thread here: https://typefully.com/dominiconorton/BNp8imI

Mainstream platforms that have launched or acquired No Code & Low Code tools

How to develop your first No Code AI application (summarization app) without coding using GPT-3 and @AdaloHQ

My no nonsense guide in a thread 🧵

First thing we're going to do is create a free @OpenAI account and make a note of our API keys.


Next you're going to need to do is create a free account with @AdaloHQ. It's a visual development tool that allows you to create native apps and progressive web applications.

You then want to create a new native mobile app starting from a blank template.

After that want to create a new screen and add a text box, button component and navigation menu.


Add a new custom action to the button.

This is going to send the input data from our native app or PWA to OpenAI to processing.

Configure the custom action settings starting with the name and type.


Following that we want to create a database table to store the response data. The minimum we need is an input, output and email column. This will store the input, output and user that generated the summary.

Make sure you add a "one-to-many" relationship to the user table.

This next part is important for the user to retrieve the results.

Add another action to the button following the custom action to create a new entry in the database you just created. Make sure you save the input, output and the logged in user.


Add another screen and place a list component on that screen.

Assign the database table you created to that list component.

Then you want to add a filter that filters the table to ensure the current items user is the same as the loggged in user.


You can add an additional details screen and action link from the list to the additional details screen to display the full information of that item on a new screen.

Click preview and you've done it! You've made your first AI application in no time.

Here's a preview of what it looks like if you're curious: https://previewer.adalo.com/48a2e78e-d1ac-4b36-8f19-b5c9b6e115d6?_gl=1*p9qqj4*_ga*MTQ0NjQ3ODQ4LjE2MzM5OTI3MDM.*_ga_SWT45DV35L*MTY2OTc4MzY3NS44Ni4xLjE2Njk3ODQ1NzAuMjUuMC4w

The most challenging part may be understanding and configuring the API request, feel free to reach out to me if you have any questions.

Every week I send No Code tutorials directly to hundreds of email inboxes. Subscribe to my newsletter and miss out on my next tutorial: https://www.getrevue.co/profile/dominiconorton


If you prefer video you can also follow me on TikTok: https://www.tiktok.com/@mr.hackathon or YouTube: https://www.youtube.com/channel/UCPYV60UNDZhM3uDciBpSh1w

How to develop your first No Code AI application (summarization app) without coding using GPT-3 and @AdaloHQ

For @AirDevCo #airhack I integrated GPT-3 with @bubble.

The project I created is called 3Word Review

User only needs 3 words and 3Word creates a complete show review.

Here's a glance at the project: https://www.loom.com/share/ff9a28a30d90476f93999711154aed03

This thread is how to integrate OpenAI with Bubble

First, you'll need to request access to GPT-3 at https://openai.com/

Next, explore some of the examples to get some inspiration and use the playground to see if the models can perform as expected for your use case.

After that, I used @getpostman to understand the API a little more and how to interact with GPT-3 via API

Then time to interact with GPT-3 via @bubble

After creating a new project, I installed the API connector in the plugins tab and created a new call.

**Note**

1. You will have to initialise the API before using 
2. Set "use as" to data initially

After that, I created a collection to store the response data.

Next, I created a workflow that triggers the API call and stores the response in my database.

**Note**

1. To use the API you have to change "use as" in the API connector to "action"
2. The API call will then be found in "plugins"

By checking in the database, I saw that the call worked with static (hardcoded) data.

Next was to add more flexibility to the call.

I added some dropdown fields and static options that allowed the user to customise their review generation.



After that, I adjusted the body of the request to use the dynamic values from the users dropdown selection.

*In the working version I go on to disable the button until the user has selected options for all the values.

Then, I tested this again and the call was working as expected and stored the customised generated review in the database.

Once this core functionality was working I begun developing 3Word using #airdev canvas, a framework built on Bubble that allows you to develop responsive Bubble applications fast. This deserved its own thread.

This is what allowed me to cut down weeks into a 2 day build

When it came to displaying the data, I used a simple repeating group.

With the aid of filters and conditions you can manipulate what data is actually seen.

Hopefully, that's straight forward for anybody that wants to create impact projects with OpenAI.

I have a newsletter that teaches you how to build software without coding, for more easy breakdowns you can subscribe here: https://www.getrevue.co/profile/dominiconorton

I also have a YouTube channel for longer in-depth tutorials that you can follow here: https://www.youtube.com/channel/UCPYV60UNDZhM3uDciBpSh1w
 

You can read the unrolled version of this thread here: https://typefully.com/dominiconorton/TZTKkTC

How to integrate GPT-3 with Bubble without coding

How to do a year's marketing for your side project in 48 Hours

TLDR: @copy_ai

Tomorrow Copy AI is allowing unlimited credits to use artifical intelligence create copy for your project.

This can be used for blog posts, headlines, ad copy and more

http://www.copy.ai/?via=dominic

How does it work?

You create an account and select create new project

I'm working on my product description for an app I have in the Google Play store but you can use it for many different use cases just find what suits you on the left panel.

Complete the desired input

Once you select create copy the tool will create various different examples for to use!

How to create content for the rest of 2022?

Usually the free tier has a credit cap but starting April 6th at 8:00 AM EST there will be 48 hours unlimited credits.

Marketing is tough but I believe over this next 48 hours copy ai will really help drive traffic for my projects. 

What are your experiences with copyai would you recommend it to a friend?