The SaaS landscape is growing faster and getting more niche by the day. With thousands of new SaaS products launching each year and categories becoming more specialized than ever, how do you keep up?

We sat down with James, the founder of SaaS Browser, to learn how his AI-powered platform is helping users discover new software, research competitors, and build connections with potential customers. From overcoming challenges in machine learning to building a fairer SaaS discovery system, James shares what he’s learned—and what’s coming next.

Why is SaaS growing so rapidly right now?

SaaS (Software as a Service) is exploding right now. Thousands of new niche SaaS businesses emerge every year, offering solutions to increasingly specific problems. Thanks to modern development frameworks, which make development easier and cheaper, cloud computing, which enables SaaS online quickly and painlessly, and open-source libraries, which reduce development time, building software is cheaper and faster than ever.

Also, with the rise of no-code app builders, founders no longer need to build software from scratch and can use an array of templates and pre-built features to get their idea up and running, without needing to write any code.

With so many new SaaS companies launching every year, keeping track of them and maintaining an eye on the industry is challenging. Usually, it's easy to discover SaaS, but only once they have become popular. This is quite sad because there are so many new and great SaaS out there that haven’t yet become popular, and it's tough to discover them.

That’s where the SaaS Browser comes in.

What is SaaS Browser, and what led you to create it?

I’ve been creating SaaS since 2009, with various levels of success. Last year was the first time I didn’t have any SaaS or any work. Usually, I’d always have the next big thing lined up to work on. It’s easy to talk about their successes, but let's take a different route and talk about a failure.

But I ended up partnering with a friend on a land investing SaaS that unfortunately didn’t work out. One of the main reasons is that we didn’t see that there were already quite a few SaaS out there that did similar things to what we built. And it was challenging to discover them, too. Unless they use SEO as a primary marketing strategy, it is difficult to find them just by searching alone. If I had a tool like SaaS Browser, it may have changed my ideas on whether to proceed with the idea because I would have discovered many more SaaS in the industry that I otherwise would have.

After that project didn’t work out, I was again stuck with the problem of what to work on next. It became clear that I needed a research tool, so I scratched my itch and started working on SaaS Browser, a SaaS directory and database to help end users discover the best software products tailored to their needs.

What is the problem SaaS Browser aims to solve?

SaaS Browser was built to address three key problems:

1. Becoming the largest online SaaS directory—helping users discover the latest and greatest SaaS products tailored to their needs.
2. Providing a powerful SaaS research tool—allowing individuals and businesses to keep an eye on competitors and emerging startups.
3. Facilitating lead generation—enabling businesses that target new SaaS companies to connect with their ideal customers.
   

How many SaaS businesses has SaaS Browser identified so far?

Initially, I estimated that the SaaS Browser would uncover around 40,000 SaaS companies. 

However, to my surprise, it has already discovered over 200,000 SaaS businesses—proving that the market is significantly larger than expected.

SaaS is becoming more niche, solving highly targeted problems that didn’t exist just a few years ago. This reinforces the need for a complete discovery platform like SaaS Browser to help users keep up with the ever-expanding SaaS industry. Developers are building apps that solve all sorts of problems, and with the rise of AI and machine learning, developers have new tools to solve even more complex issues. Even at SaaS Browser, we also use a healthy dose of machine learning (more on that next).

How does SaaS Browser work?

At the heart of the SaaS Browser is a classification AI model that determines whether a website is a SaaS business or not. The model assigns each website a score between 0 and 1, with higher scores indicating stronger SaaS characteristics. When a website has stronger SaaS characteristics, it's added to our index.

What challenges did you face in training the AI model?

One of the biggest challenges was obtaining quality training data. I needed thousands of labelled examples—both SaaS and non-SaaS—to train the model effectively. While I initially outsourced the data labelling process, the results were inconsistent, even after providing extensive training materials and video demos.

To solve this, I developed a custom tool that allowed me to open and scan 100 sites simultaneously, enabling me to label about 1,000 sites in just a few hours. I even invested in a new MacBook Pro with 48GB of RAM to handle the heavy browser load.

Iterative Approach to Machine Learning:

The model was built using an iterative approach. First, I trained an initial version of the model. Once trained, I classified a large batch of websites. After classification, I manually reviewed the sites the model was unsure about. Any corrected classifications were then added to the training data. Finally, I retrained the model and repeated the process to improve accuracy.

One challenge that emerged was differentiating SaaS from marketing agencies. Since many SaaS products are marketing-related, the model frequently misclassifies agencies as SaaS. Future iterations will involve adding more labeled examples to improve accuracy, since sadly, some marketing agencies are included in the directory that aren’t strictly SaaS.

How did you manage cost optimization?

Training AI models can be expensive, especially when using GPU-based computation. To keep costs under control, I opted for Hugging Face Endpoints, which automatically shuts down when no inferences are being made, significantly reducing cloud computing expenses. That being said, with tens of millions of active websites on the internet, checking each of them is still quite expensive, and it took several weeks of computation to discover which ones were SaaS.

What about categorizing SaaS—how granular is it?

Another major challenge was deciding how wide or narrow to categorize the SaaS. For example, is a SaaS a marketing tool (quite narrow) or a marketing tool that helps PPC optimization for Facebook Ads (quite wide)? 

A wide categorization system offers highly detailed classifications, making it easier for users to pinpoint SaaS solutions that precisely match their needs. However, the tradeoff is accuracy, since the deeper the classification goes, the fewer training examples exist for each subcategory, leading to potential misclassifications. 

On the other hand, a narrow categorization system is more reliable, as it has a larger data set per category, resulting in higher confidence in classification. However, it may not offer the specificity some users desire when searching for a very niche SaaS.

Hitting the right balance between depth and accuracy has been difficult. After testing different approaches, I ultimately settled on a hybrid approach that leverages open-source AI models combined with keyword-based classification, using around 160 different categories. This allows for a structured yet flexible categorization system, which can be fine-tuned over time as more training data becomes available. 

In any case, using humans to categorize each SaaS would have been too time-consuming and very expensive, since SaaS Browser has over 200,000 SaaS, so using an automated way to categorize SaaS was absolutely necessary.

What lessons did you learn building SaaS Browser?

One of the most essential takeaways from building a SaaS Browser is that an iterative approach to AI training is critical. Machine learning models are only as good as the data they are trained on, and refining that data through repeated cycles of testing, feedback, and retraining significantly improves accuracy.

Moreover, machine learning is an absolute necessity due to the sheer number of SaaS businesses out there. Manually classifying over 200,000 SaaS websites is simply not feasible without automation, and AI plays a crucial role in making this classification scalable and efficient.

Despite the power of AI, human expertise is still indispensable. While machine learning models can process vast amounts of data quickly, they still struggle with nuanced distinctions, such as distinguishing between a marketing agency like Dayta and a true SaaS product. Human intervention is required to curate high-quality training data, correct errors, and refine classifications over time.

A key lesson has been that no AI model is perfect from the start. The best approach is to start with a baseline model, identify misclassifications, improve the dataset, retrain the model, and repeat the process. Over time, this cycle leads to a more reliable and accurate system.

What has the response been since launch?

Since launching, it has become clear that there is a strong demand for a platform that helps discover smaller, lesser-known SaaS products. 

Many startup founders and businesses struggle to get their SaaS products in front of the right audience simply because they don’t have the marketing budget that larger competitors do. 

Traditional platforms like Capterra and G2 primarily operate on a pay-to-play model, meaning only well-funded SaaS companies gain visibility. This creates a skewed ecosystem where emerging yet innovative SaaS products often go unnoticed by potential users who might benefit most.

The problem is that thousands of SaaS companies are being built to solve highly specialized issues. Still, these solutions are difficult to discover without a central database not influenced by advertising spend. 

SaaS Browser is changing that by providing an unbiased, data-driven discovery platform that allows users to find and research SaaS solutions based on their actual value and functionality rather than their marketing budgets. 

By cataloguing a vast number of SaaS businesses and providing real-time insights, SaaS Browser opens access to SaaS visibility, giving smaller companies a fairer chance to discover and succeed.

What’s next for SaaS Browser?

Moving forward, SaaS Browser will focus on several key improvements to enhance its usability and effectiveness:

• Introduce more powerful filters and insights—users will soon have access to advanced filtering options, allowing them to refine their searches based on industry, pricing models, feature sets, and business size. Additionally, new insights will be provided, offering a deeper analysis of SaaS industry trends and patterns.
• Provide trend data on fast-growing SaaS—using data analysis and machine learning, SaaS Browser will track and highlight SaaS businesses experiencing rapid growth. This will help users identify emerging market leaders and up-and-coming competitors in near real-time.
• Enhance categorization for better discovery—improvements will be made to the AI-driven categorization system to provide more precise and granular classifications. This will make it easier for users to navigate the SaaS ecosystem and find what they need more accurately. New categories of SaaS are appearing all the time. For example, who knew that an AI avatar SaaS would be created just 10 years ago? This will be an evolving challenge.
• Develop a reporting system to refine accuracy—a feedback loop will be implemented, enabling users to report incorrect classifications or outdated information. This data will then be used to retrain the AI models, ensuring continuous improvements in accuracy and reliability.
• Improved SaaS profile pages, including videos of products, etc.

One of the most exciting opportunities for SaaS Browser is the potential to license our SaaS data feed to businesses and researchers who need comprehensive, up-to-date information about the SaaS industry. With a growing database of over 200,000 SaaS companies, we are uniquely positioned to provide invaluable insights into market trends, emerging competitors, and new software categories. 

Also, plenty of investors are out there looking to acquire SaaS companies. I know this very well from my experience working with Empire Flippers and WebStreet, and they have strict buying criteria, so it will be interesting to work with other companies.

Discover amazing SaaS products today at saasbrowser.com. For feedback or inquiries, reach out at james@saasbrowser.com.