I Grew My Paraphrasing API To $3.5K/Month [$0 Spent In Ads]
Hello! Who are you and what business did you start?
Hi, my name is Boulama and I’m the founder of Tinq.ai.
Tinq.ai is a ready-to-use NLP toolkit that allows the integration of Natural Language Processing capabilities into your app.
It is designed to be an intuitive and easy-to-integrate platform to assist you in producing automation solutions for tasks related to natural language processing.
Tinq.ai has several very popular products amongst our community: rewriter, sentiment analysis, entity recognition, and classifier. It was launched in private beta in July 2021 and has been accessible to the public since October 2021. The platform recently crossed the $3.5k revenue mark with 34 paying customers as of January 27, 2022.
What's your backstory and how did you come up with the idea?
I have been an entrepreneur for a long time I started my first software development company at age 18 when I was living in Niger (West Africa), a company that I still own today. This position gives me a lot of insights into what business owners need to make their businesses more productive.
Ever since the beginning of the pandemic in 2020, I started noticing the need for businesses to automate as much as possible to remain competitive and productive.
Many of my software business’ clients are in this category asked to build tools for their customer service & IT departments to process clients' requests faster.
I have always been a big fan of machine learning, especially NLP (natural language processing), and I saw this as an opportunity to learn more and eventually build something in the AI space.
I searched a little and found a couple of platforms doing the same but most did not have a simple “plug and play” type of API. And when they did, the price was atrocious. I decided that it could be something that I could try to take on.
I first built a simple paraphraser, summarizer, and classifier using open source models built by Google, that performed decently well. I fine-tuned them and I was able to pitch some of my clients with that original idea that I could build even more for them for a fraction of what it would cost if we were to build custom solutions for them.
Many have been romanticizing starting a business lately, especially the failure part of it. But I believe that with good preparation and clear planning, a lot of failures could have been avoided.
I am very familiar with API development, so I came up with a rough prototype fairly easily a couple of weeks later using Flask, Amazon Sagemaker, and Laravel. When I presented it to my potential users they were excited that they didn’t have to pay thousands of dollars to have such services, and we would just implement the API in their workflow. And that’s how I validated the idea.
Take us through the process of designing, prototyping, and manufacturing your first product.
Once I talked with my initial customers (those who helped me validate the idea), I started building something that would be easy enough to use even for nontechnical people. I had a very vague idea of how machine learning works but had never built or fine-tuned anything so I had to learn.
Again, thankfully I had a lot of credits from cloud providers that I used to have access to powerful GPUs to fine-tune models for the specific tasks that were needed for the platform such as summarization, paraphrasing, classification, entity recognition, and much more.
At first, there was no particular user interface or anything like that the users would just sign up and have their API credentials and a basic request to get them started with the endpoints and a gauge diagram that would show users how much of the API they had consumed.
I spent a lot of time discussing with the first set of users and gathering their feedback to design a very simple UI that could be used. After a couple of weeks and adding more features, a shy UI started to emerge which is the one that we currently have now on the site.
This is what the dashboard looks like today:
Describe the process of launching the business.
One interesting thing about Tinq.ai is that there was never an official so-called launch. I would come now and then on Twitter, where I have a decent size audience (with about 5.7k followers), and just share the link with an update. As I said in the previous paragraphs, there was already a set of users that I was testing Tinq.ai with; they did share it within their community and we got started from there.
However, around November, there was a plateau in the number of sign-ups that were getting. I then decided to learn a bit more about SEO and optimize that aspect. Thanks to that, most sign-ups that now get at Tinq.ai come directly from Google and the most popular search is “paraphrasing API” for which we are always on the first page!
I started to be good at this game and in mid-January 2022, I decided to buy the domain rewrite. app, on which I built a free version of the main paraphraser available on Tinq.ai. After just 3 days it was a hit and it became a huge source of traffic for Tinq.ai (screenshot below).
Since launch, what has worked to attract and retain customers?
When launching the product, I didn't have a clear strategy of how to do it and how to get new users. The only thing that I did was just to get the product out there and see what would happen. Until this day, I have spent zero dollars on advertising, so all the growth is purely organic.
There are a few things that, in my opinion, work great to attract and retain customers.
The first thing I did when launching the site publicly was install Crisp.chat, which is a live-chat tool to interact with users that would come on the website. And I would get a push notification anytime someone opens Tinq.ai’s website.
Users who love your product will HELP you improve it!
Using that, now and then I’d randomly send a message with something like “hey how can I help you” or “is there anything you’d need help with”.
Usually, people would respond and ask questions, and most of the time they would end up purchasing a paid plan. This made me realize how important it was to discuss with users. Now there’s a full-time customer support rep that takes care of that, but the recipe is the same, let’s talk with users and help them in any way that we can.
The second thing is a direct consequence of the first one, which is using user feedback to build new features in existing products. For example, when I first launched the classifier, to create classifiers, users had to manually enter their labels. Now, by studying the most commonly created labels, Tinq.ai has a template that allows users to create custom classifiers from templates, and they can modify or add new labels as they wish.
How are you doing today and what does the future look like?
Tinq.ai is still very young and in its early stage. A lot is coming in the future in terms of features and growth. I would not say that it is profitable yet. Training AI models requires a lot of computational power and we’re currently using free credits to do so. Once they run out, that needs to come out of pocket. But it recently passed the $3,400 MRR (excluding initial customers who helped validate the idea). And most of the growth happened in January 2022, alone!
I hope to reach $10,000 MRR soon and start promoting the platform through various channels. As I said previously, so far I have not spent anything on ads and I hope to keep it this way for as long as possible. I am not a big believer in traditional ads, I think that your users are the best ambassadors of the products that you’re building. Loyal users do spread the word about products they love for free already, and I think that it is a decent incentive to offer a little reward for them to promote the tool they love to their audience.
Nothing beats word of mouth, in my humble opinion. Therefore, the next step is to launch a community and start affiliate marketing with the current most loyal users.
Through starting the business, have you learned anything particularly helpful or advantageous?
Tinq.ai is not my first business but it is the first real SaaS that I launched. But so far, the amount of learning that I do daily is insane.
If there’s anything that I’m learning with this all SaaS thing, it’s that talking with your users constantly is key to gaining trust and loyalty. What you’re building can be average, or just good enough.
Users who love your product will HELP you improve it!
When it comes to timing, COVID 19 Pandemic made many businesses realize the power that automation and AI can bring to their businesses, and I am lucky enough to have a software company with the insights necessary to understand what can and cannot work.
Since starting Tinq.ai, I must admit that I am addicted to Google Analytics and Hotjar. The latter is just godsent. It allows us to have so much information about how users interact with a product, and how we can tweak the tool to make it even more appealing to future and current users.
What platform/tools do you use for your business?
When it comes to tools that I use, there are plenty. But here are the most relevant when it comes to building Tinq.ai.
- Crisp: Live chat system.
- Airtable: Whenever a user unsubscribes from Tinq.ai, they have displayed a survey and the answers from this survey go to an Airtable database.
- AWS.com - Sagemaker: Deploy AI models.
- DigitalOcean is my go-to hosting platform.
- Google Analytics - Website analytics.
- Hotjar - User behavior analysis tool.
What have been the most influential books, podcasts, or other resources?
I am an avid reader and consumer of startups oriented content.
It might’ve been mentioned here but one book that I’ve read that helped me tremendously through building Tinq.ai is Zero to One by Peter Thiel. And another great one that has influenced my entrepreneurial journey is Principles by Ray Dalio.
When it comes to podcasts, there are two podcasts that I never miss, the first one is This Week In Startups by Jason Calacanis and the second one is How I Built This with Guy Raz. These, I believe, are the best resources for builders or people who want to get their feet wet.
Advice for other entrepreneurs who want to get started or are just starting out?
There are two pieces of advice that I can give to anybody who’s getting started or thinking about starting a business, no matter if it is in technology or not.
The first one would be, have a clear plan of what product you want to build, which also implies that you need to know the industry that you’re getting yourself into.
Many have been romanticizing starting a business lately, especially the failure part of it. But I believe that with good preparation and clear planning, a lot of failures could have been avoided.
That leads me to my second point, which is being focused on delighting customers as much as possible, it can be with great customer service with a great tool, too. Try to engage with them and make it easy for them to reach out. Like Tinq.ai has a live chat available on each page.
To conclude, no matter which product you’re building please treat your customers and users with the highest level of respect possible.
Where can we go to learn more?
If you have any questions or comments, drop a comment below!
- 4,818 founder case studies
- Access to our founder directory
- Live events, courses and recordings
- 8,628 business ideas
- $1M in software savings
- 4,818 founder case studies
- Access to our founder directory
- Live events, courses and recordings
- 8,628 business ideas
- $1M in software savings
- 4,818 founder case studies
- Access to our founder directory
- Live events, courses and recordings
- 8,628 business ideas
- $1M in software savings
- 4,818 founder case studies
- Access to our founder directory
- Live events, courses and recordings
- 8,628 business ideas
- $1M in software savings
- 4,818 founder case studies
- Access to our founder directory
- Live events, courses and recordings
- 8,628 business ideas
- $1M in software savings
- 4,818 founder case studies
- Access to our founder directory
- Live events, courses and recordings
- 8,628 business ideas
- $1M in software savings
- 4,818 founder case studies
- Access to our founder directory
- Live events, courses and recordings
- 8,628 business ideas
- $1M in software savings
- 4,818 founder case studies
- Access to our founder directory
- Live events, courses and recordings
- 8,628 business ideas
- $1M in software savings
