It's June 2024, and Generative AI needs no introduction. From simple cool demos at the beginning of 2023, it has evolved to a state where companies are beginning to see measurable benefits from Gen AI. These benefits range from a simple increase in productivity to improved customer experiences that result in revenue growth. Billions of dollars have been invested in this sector. Hundreds (if not thousands) of AI companies are born every month, and interestingly, many prominent SaaS companies are pivoting to become AI first. And here we are, starting yet another AI company, and that too with a service-first approach.

So, the only question is - What problem are we going after and why?

To help you understand that, you must first understand what types of AI companies exist.

The above is an oversimplified way to categorize these companies. For a more detailed version, read this post by Nfx.

But staying on the topic, we are here to solve problems for categories 1, 2, and 3. For them, AI is the catalyst for growing their business, not their core business itself. Most companies in categories 1, 2, and 3 need help to tap the full potential of AI.

The real problem is - How to improve the bottom line using AI

According to recent reports, despite so much progress, only about 5% of GenAI projects lead to significant monetization. Infact, some estimates claim that 85% of the time, companies are investing in GenAI projects without a clear understanding of how they will use the technology to create business value. And the reason for that is straightforward - AI alone is not the solution!

In many cases, it's a pivotal but only a tiny part of the most optimized solution. This is where we exist. We start from scratch by defining the problem first. This step is critical as it helps our clients determine whether their business needs Gen AI. In many cases, simple machine-learning techniques and rule-based algorithms are the right solution. After that, we take an iterative approach towards designing and executing the solution and bringing it to production.

Why now?

Another dimension of our decision was why this is the right time to start an AI company. For us, the answer is straightforward. We have been experimenting with Gen AI since the beginning of last year, and for the first time, we can see tangible results that can create real business value beyond writing a few copies and generating a few memes. Gen AI offerings have seen a meteoric upgrade. And I am not only talking about the capabilities of the LLMs. Taking AI to production has become much easier in comparison because of tons of open source and enterprise offerings around it. Even experimenting with AI is relatively easy now. The AI models and ecosystem have seen a massive upgrade and continue to show promise in different domains.

If I have to draw a parallel, Gen AI has grown from a toddler to a 10-year-old kid. Imagine what will happen when this kid comes of age!

Hallucination, high cost, and high latencies are the symptoms, not the disease!

Like everyone else, our first thought was to start with a SaaS business. However, once we started taking our AI agents to production, we knew that one-size-fits-all solutions would not work with Gen AI. The first reason is simple—it's your company's data!

In an ideal world, every Jira ticket should have a full description, every sales associate should update the CRM after their call/meeting, every line of code written by the developers should be super optimized, and every customer support agent should respond to customers with entirely accurate information and every product/engineering/process/management document written should be pristine.

However, anyone who has ever run or worked in an organization knows and accepts that despite the best efforts, the data generated and stored by the organizations can be in better condition. And this data is the input and context of Gen AI.

The quality of the data is one problem. Other factors, such as the data pre-processing technique used, the Choice of LLM, the Choice of your Vector database and file stores, the Choice of embedding model, and the architecture design of your Gen AI production setup, can affect the results.

So, the most optimal and practical solution has to be tailored to the organization's data, context, and stage of life. It must be married to your go-to-market strategies, team capabilities, and future business goals.

Gen AI is just getting started, and so are we

Through a lot of luck and some good decision-making, We have been at the intersection of Product, Growth, and technology in the last decade. We have been part of multiple zero-to-one journeys and have scaled technology systems from zero to millions of users and millions of dollars in revenue.

And we know how to design solutions that work!

Now that we have answered the "what" and "why," you can ping us here to find out the " how"!