Issue#18: Building $1K-$10K MRR Micro-SaaS AI/ML products using AWS/GCP Cloud services
We are excited to share the latest Micro SaaS news and Micro SaaS Ideas to help you on your journey as a SaaS founder. Let's explore what's new this week!
👉 Get your Pro subscription today - Get access to hundreds of Micro SaaS Ideas, updates on Micro SaaS trends, and access to a closed community of founders, builders and makers.
Unlock the Pro version
Get the full version of this newsletter and all previous letters by subscribing to Pro version.
No fluffy content. If your goal is to build a $100m ARR business, this is not the right post. Here I am are NOT going to talk about building the next Facebook or Twitter. If your goal is to make a $1K to $10K MRR, continue reading.
This post will cover one SAAS area and talk about multiple niches in this space. This post also explains more on how to do tech implementation, do market analysis, how the current players are doing, and also ends with a cost analysis to understand the overall cost for 100 users.
Throughout this post, we will be talking about various terms. Here is a quick explanation of various terms we talk about.
AWS - Amazon Web Services - A cloud platform by Amazon
GCP - Google Cloud Platform - A cloud platform by Google
Microsoft Azure - A cloud platform by Microsoft
AI - Artificial Intelligence
ML - Machine Learning
Let's see some of the Micro-SaaS products built using AI/ML and see how well we can build new products in this niche without using any infrastructure or monthly commitments. The goal of this post is to show the possibilities to build products around AI/ML by just using cloud APIs without worrying about the complexities and spending huge money.
These terms like AI/ML often scare people and in the recent past with the advent of cloud services, build products related to AI/ML has become extremely easy. So, if you are new to AI/ML, don’t worry and we will dive deep and see the various possibilities here.
Current players
ConveyThis: Modern Language Translation Switcher for websites with 5 minute instant setup. Currently making $10K MRR.
Get Spread: Easiest way to transcribe. Made for pro transcribers. New player.
Bablic: Bablic makes managing setting up global websites simple. No hassle, no coding, just point & click to set it up and manage updates. Choose machine or human translation. Currently $3.3M annual revenue.
Talkwalker: Combine leading social media analytics with the industry's most comprehensive visual listening technology
DatatoBiz: Improve your business efficiency and real time decision support with advanced Computer Vision Solutions
Imagga: Build the next generation of Image Recognition Applications with Imagga's API . Founded in 2008. $10m in revenue.
Visive:AI Powered Visual Automation for Business.
Brand24: Internet monitoring tool combined with advanced sentiment analysis working together to protect brand, analyze audience opinions, and connect companies with potential customers.
TextCloud: First automation platform capable of understanding text. Load data from sources like Google Sheets, add sentiment analysis or other language processing modules and trigger actions based on the results.
Trint: Audio transcription software. Founded in 2014. Estimated revenue $17m/year.
Zpoken.ai: Get your audio and video transcribed in minutes. Founded in 2017. Currently at $15K MRR.
Spext: Looks like a doc. Is actually an audio editor. Edit the audio by editing the auto generated transcript, add music & merge recordings together. Founded in 2017. Raised $300K in funding.
Jotengine: Automate transcription with Jotengine Transcription and Caption API.
Valossa: Most Advanced Video Analysis Software . Founded in 2015. Making $10m in revenue.
Negative Nancy says - “AI/ML is a complex term and needs years of research and time to build something in AI/ML domain”
Me - AI/ML was a complex term. But cloud services have made it easy to build products around AI/ML. The core AI/ML engine is wrapped up in cloud services in the form of APIs.
Negative Nancy says - “AI/ML needs a lot of technical knowledge to build products”
Me - If you know how to call an API, you can build cool products around AI/ML using AWS/GCP/Azure.
Negative Nancy says - “Why not just use GPT-3 instead of AI/ML using AWS/GCP/Azure cloud services?”
Me - Well, You can use GPT-3 also. Don’t underestimate the power of GPT-3 but that is still in beta and not available for everyone. There are still thousands of people waiting for GPT-3 access. On a side note, GPT-3 cannot do everything that current cloud services could do - for example - “Image recognition”, “Building Search engine”, “Building chatbot” etc are not directly possible using GPT-3. On top of that, GPT-3 costs you at least $100/m.
Deep-dive
The one thing that is common in all the above players is that they are all built on top of some amount of AI/ML. The very word “Artificial Intelligence” may be scary but everything that we will see in this post can be built just by calling APIs.
For example, a service like “Text Translation to different languages” can be just achieved by calling simple AWS APIs or GCP APIs. You can build products similar to ConveyThis (make a note that ConveyThis is making $10K MRR.
The beauty of using these cloud services is that - all these work on a “pay-as-you-go” model and there is no upfront payment or monthly commitment required. For example, if you build something and are worried about any upfront costs when no customers use your product, don’t worry - You would be paying only based on the usage.
Scales automatically - Another major thing that cannot be ignored. The biggest advantage with building Micro-SaaS products around these cloud AI/ML services is that they scale automatically based on usage. Whether you use these of 1 API call or 100K API calls, they scale automatically without you provisioning any infrastructure. If you never worked on things at scale, you may not be able to relate to the pain points but if you ever worked on products that spike to thousands of transactions per minute, it is extremely important to see how well your architecture scales.
I didn’t mention Azure here as that could overwhelm me. But most of these can be achieved using Azure as well. If you are keen on using Azure to build any of these ideas, that should work too.
Some niches
Extract metadata from Images: An automated solution that can recognize the objects from an image. There is a lot of market around this and cloud APIs can detect the objects in the image (if an image has male, female, child, object, hill, cycle, mountain, etc), the mood of the objects in the image (if people are sad, happy, celebrating), dominating colors of the object and a lot of other data as well. If you are wondering how this looks like or how it would work, goto Google Vision cloud service and upload an image (requires no login and code to test) and see how it works. You can create a Micro-SaaS product around this functionality. There are various use cases for this - See how Amazon Rekognition can be used for workspaces safety. Also, see how you can use this for content moderation. This can be also used by e-commerce companies to extract the colors from the images and show related products on the e-commerce store.
Building Chat Bots: Contact forms are dead long back. People love chatbots. Now our SaaS tool shouldn’t just make simple bots. The bots should be intelligent enough. The bots should be able to reply to the user by querying the database, looking at a file, querying an API. This sounds complex, right? But with Amazon Lex, a chatbot solution, this is super easy and you just have to learn a few things like calling backend API using AWS Lambda. Alternatively, a similar solution can be developed on GCP as well using GCP Dialogflow and on Azure with Azure Bot services. You can start this as productized service as well instead of creating a SaaS out of this. This will also help you understand user’s core problems and requirements before you build a SaaS.
…. get another 10+ additional niches from Pro version.