The Challenges of Implementing AI in a SaaS Giant | Typeform Podcast

In today's landscape, the way we collect and interpret data is undergoing a seismic shift, driven by advancements in AI and user experience design. According to Salesforce, 91% of business buyers said the experience a company provides is equally important as its product offering.

One company at the forefront of improving user experience through forms is Typeform, a no-code SaaS platform that has transformed the mundane task of filling out forms into an engaging, conversational experience. Known for its user-friendly design and innovative features, the company has recently ventured into AI-driven data collection through its new product, Formless.

Our host Vianca Meyer sat down with David Okuniev, the co-founder of Typeform, to discuss the importance of balancing design and data and how AI is shaping the future of data collection.

Who is David Okuniev?

David is the co-founder of Typeform who leads the R&D team as a product designer in charge of bringing new products to market. His mission to make digital engagement a little more human resulted in the creation of Typeform in 2012. With a focus on interaction, UX, and UI, David and Typeform's other co-founder Robert Muñoz sought to create a platform that would enable brands to interact with customers in a dynamic, conversational way.

This podcast transcript has been edited for clarity and readability.

Vianca: What sparked the idea to launch a no-code SaaS platform such as Typeform back in 2012?

David: Over a decade ago, we stumbled into creating Typeform without any prior knowledge of launching products or understanding SaaS metrics. Our background was in agency design work, and we initially started Typeform as a side project to make forms more engaging. We didn't set out with the intention to build a no-code platform; it just evolved that way. Over time, we realized that our tool became incredibly flexible and could be integrated into various workflows. We've even seen entire businesses use Typeform as their front-end interface.

What were the biggest challenges in transitioning from a small-scale web development agency to Typeform?

The early years at our company were exhilarating, marked by rapid growth and a close-knit team that felt like family. However, challenges emerged as we scaled the company. We began to see internal factions and a communication breakdown, which led us to implement more structured processes and accountability measures.

To address these challenges, we emphasized the importance of a clear mission, vision, and values. We consistently communicated these principles to keep everyone aligned. One of the biggest issues in scaling is the creation of silos and loss of communication. To counteract this, we made efforts to keep communication channels open and everyone on the same page. The key to successful scaling, in our experience, has been maintaining a unified focus and direction for the entire team.

Typeform's conversational approach to forms is pretty distinct. Why do your users love it so much?

I firmly believe that the most effective way to collect information is through a conversational approach that engages the user. We have data to back this up; tests comparing Typeform to standard forms show significantly better conversion rates. On average, Typeform has a 60% completion rate, which varies depending on the use case.

Our approach combines good form mechanics, like asking one question at a time, with a well-designed user experience that includes elements like images and large typography. This creates a more engaging, presentation-like experience rather than just a list of questions. We've found that this approach fosters better connections between B2B companies and their customers. We call this "experiential data collection," and we use design as a tool to achieve it. We're always striving to stay ahead of the curve in this area.

How did you keep user experience and design at the heart of Typeform as it grew and diversified?

Scaling a design-oriented, emotionally-driven company has been a challenge. As we grew, we had to balance the creative, impulsive types with the more rational, data-driven individuals. Finding the perfect blend between heart and mind hasn't been easy. At one point, we overcorrected toward rationality, and we've spent the last few years trying to bring the balance back to the center.

Being design-driven means sometimes making decisions based on instinct and vision, not just data. If you only act on fully measured and validated ideas, you risk losing speed and innovation. We're working to embed this balanced approach into our company culture. We're formulating values that reflect being both "driven by design" and "data-informed," emphasizing that while we value design highly, we also commit to thorough analysis and optimization.

In May, you introduced an AI-driven form builder. Can you take us behind the scenes of its creation?

When we first built Typeform, we aimed to make it as conversational as possible, inspired by the idea of interacting with an AI. We included features like logic and branching to make the experience feel more like a real conversation. Now, with advancements in AI and large language models (LLMs), we're exploring how to make forms fully conversational in real time through Typeform Labs.

Our new approach allows the form to react dynamically, offering comments, quips, or jokes based on the user's input. Unlike the deterministic flow of traditional Typeform, this new form is open-ended. You provide a prompt and specify the key information you want to collect, and the AI handles the rest.

We've even added a feature where you can train the AI with your own data, like your website URL or help docs, so it can answer questions about your company during the form interaction. We're currently in closed beta but plan to move to open beta soon. This represents a significant evolution from what Typeform currently offers.

What were the major challenges in implementing this advanced AI into Typeform's existing products?

In Typeform, we've integrated AI to streamline various aspects, from form creation to data analysis. However, with our new product, Formless, AI is not just an add-on; it's the foundation. When you interact with a form on Formless, you're essentially conversing with the AI. This is different from Typeform, where the interaction follows a predetermined script.

The main challenge has been working with large language models (LLMs), specifically GPT-3.5, to tailor them to our needs. Getting the prompting right for various use cases has been technically challenging. While language models are powerful, their capabilities are still limited in some respects, such as avoiding hallucinations in responses. However, as they continue to evolve, like the upcoming GPT-4, we expect it will become easier to build more advanced and reliable products like Formless.

How do you see AI fitting into the future of data collection and human-AI collaboration?

The purpose of Formless is to innovate and stay ahead of the curve. When asked if Formless will replace Typeform, the answer is uncertain and largely depends on how user experiences and market demands evolve. Right now, people are accustomed to using forms in a traditional manner, but who's to say that will be the case in five or ten years? That's why we're developing products like Formless that look to the future.

If Formless does become the new Typeform, it will be because the market has shifted in that direction. Currently, the majority of our nearly 200,000 paying customers are using Typeform, which offers a wide range of features that Formless doesn't yet cover. So, while it may take some time for a transition to occur, we're committed to staying ahead of the curve and meeting the evolving needs of our customers.

Are there any upcoming projects at Typeform that you'd like to tease our readers with?

We're working on a feature that allows users to interact with their data in a conversational manner. For example, you could ask, "What are people generally saying about the quality of my product?" and receive natural-sounding answers in return. This will be integrated into all our products, including Formless. The idea is to make it easier for users to get an overview of their data without having to sift through each response.

Recently, we introduced a feature on Typeform that enables users to create forms simply by providing a prompt. For instance, you could say, "Build me a lead gen form for my beauty clinic," and the system will automatically construct the form for you. Formless operates on a similar principle, but it's more open-ended. You provide a prompt, and the AI takes care of hosting the conversation, making the process even more dynamic and interactive.

Edited by Nikola Djuric. Keep up with Typeform and check out more of our interviews with industry experts!