AI Amplified: The Art of Customer Experience

This time we’re introducing our readers to a transcripted version of a stage talk at TechBBQ Copenhagen, with our partner Pauliina Martikainen-Rahnu and Magic Feedback Co-founder and CEO Nima Vali Rajabi.

Following their fireside chat on stage, Nima received a wave of praise from founders and investors to his inbox, saying this was one of the most compelling conversations the Gen AI landscape has seen in the Nordics, which is why we are allowing a wider audience to relive it through the transcripted version.

Pauliina: Hi all and thanks for joining us! We have a very customer experience and AI-focused 25 minutes ahead of us. And as I know that you want to get to the actual topic, I’ll keep the introductions short. I am Pauliina Martikainen, Partner at a Nordic VC firm Maki.vc, out of which we invest in pre-seed and seed stage companies building deep tech and brand-driven companies. And with me, I have Nima Vali Rajabi, CEO and Co-Founder of Magic Feedback, a company turning customer feedback into insights with AI. Maki and our case team, me & Caroline Gattner, had the privilege of getting to lead Magic Feedback’s very first investment round and it’s been such an exciting journey since, working with the Nima in this space that’s really just demanding better solutions.

Nima and Pauliina on stage at TechBBQ

The Role of AI in Customer Experience

Pauliina: Nima, an opening question to kick this off with and to paint the broader question: AI is rapidly transforming the way businesses interact with their customers. What are the most significant advancements generative AI has brought to customer feedback management?

Nima: One of the hardest challenges in customer feedback management, that no one has been able to crack yet, is how to manage customer feedback at scale. It’s easy to keep track of things when you’re a small business, but it’s impossible to do so in a large company when you receive tens of thousands of online reviews, support tickets, sales conversations, and whatnot. A lot of really smart people have tried to create processes and find ways to solve this problem. I’ve also spent the last decade trying to solve it.

Even at Google, where I used to work with some of the smartest people and best technology available, we just couldn’t crack it at scale. However, now, for the first time, with generative AI, we can analyze unstructured customer feedback at scale with human-level accuracy. That means we can finally solve the problem of how to manage customer feedback at scale without any manual processes.

Up until 2023, all the software we used to manage customer feedback was built within the technological constraints of the past decade, where it was impossible to make sense of the data that went into these systems. If you log into any software, you have to manually tag, categorize, filter, and export things. But with generative AI’s ability to make sense of unstructured data, we now have a once-in-a-lifetime opportunity to reimagine and reinvent all the software solutions in the space. Most importantly, we can now solve many of the problems that have historically been impossible to address at scale.

Implementing and scaling AI internally to promote growth

Founders of Magic Feedback Nima Vali Rajabi and Francisco Arias

Pauliina: Many companies want to implement AI solutions yet struggle to scale them internally. There are plenty that work with customer experience and have told you that their CEO is asking them “why they are not doing anything with AI” and they are almost desperately looking for concrete solutions to do so at scale. What do you believe is the key to successfully integrating AI into company strategy to support rapid growth, as you've done with Magic Feedback?

Nima: As counterintuitive as it might seem, our job is to rock the boat as little as possible. What I mean by that is, that even though we have the opportunity to reinvent and reimagine all the software solutions in the space, we still have to do it in a way that introduces as little friction as possible into companies’ existing processes and performance.

A good example is our Magic AI surveys. We know that survey open rates and completion rates are vital to our companies, so we built AI surveys that look and feel exactly like the companies' existing surveys, only with the option to ask really good AI follow-up questions when there’s an opportunity to collect better and more actionable feedback. Many of our enterprise customers send millions of surveys a year, and people's bonuses are tied to the scores in those surveys, so we can’t just introduce a new survey format that creates turbulence and impacts the company's survey performance without good reason.

While it’s tempting to make AI front and center with a lot of noise around it, it’s much more important and challenging to show restraint in terms of respecting the users and opportunities when we innovate.

Integration & handling of enterprise-level feedback data

Pauliina: Still on the same theme, a common challenge that many companies face is integrating new AI tools with their existing systems. How should startups approach this in your opinion? How does i.e. Magic Feedback address integration challenges?

Nima: Right now, we are in the transition from legacy SaaS software to AI-native software, and that transition requires us to be the ones to build the bridge between the old and new worlds of SaaS. We can’t just replace legacy SaaS or expect people to know how to bridge the two. All of our enterprise customers look to us for help and guidance on how to do this. So it’s been essential for both us and our customers to learn how to build that bridge.

This means we do everything humanly possible to make our customers successful with AI, from conducting in-person workshops on AI best practices to creating custom integrations with systems that are almost impossible to integrate with. In many ways, you need to have a foot in both the old and the new worlds if you want to bridge the gap between them.

We often come into companies that have been running a POC with an AI company for months without any progress, and we almost always see the same pattern of AI startups failing to build the bridge between the two worlds. As obvious as it might seem, it is really complex and challenging to effectively build that bridge as an AI startup working inside an enterprise company.

Nima Vali Rajabi, Co-Founder of Magic Feedback, Pauliina Martikainen-Rahnu, Partner at Maki.vc and the beautiful setting at the TechBBQ

Ethical & security considerations in AI-driven feedback

Pauliina: One thing that needs to be addressed early on in our chat, too, is security and ethical considerations. This is vital for all companies and the bigger the prospect the higher the bar to satisfy their needs. As AI becomes more involved in gathering and analyzing sometimes also sensitive customer feedback, how do you ensure that the insights generated are free from bias and truly reflective of diverse customer perspectives?

Nima: Our philosophy from day one has been that you should always be one click away from transparency into how the AI works. In practice, that means you can click on anything our AI generates on our platform and see all the original data that supports the AI’s work and how it decided to analyze it. This was a lot of work to implement, but one of the key lessons we took from Google is that when you fully automate something and ask people to give up human control over manual tasks, it’s essential to provide as much transparency into the process as possible.

Pauliina: Can you talk a little bit about that? What is, in your opinion, the best approach to safeguard sensitive data?

Nima: Before we built Magic Feedback, we asked ourselves if we would feel comfortable copying and pasting all our sensitive data into ChatGPT and hitting submit. And as much as I love ChatGPT and use it every day, we definitely did not feel comfortable doing that. So if we didn’t feel comfortable doing it, how could we ask our customers to do the same?

That meant we took a much harder path than most AI startups, as we decided to build our own AI models on our own servers. Although it was one of the hardest technical things we’ve ever done, we are proud to say that our customers’ data never leaves our servers to be analyzed by OpenAI or any third-party providers.

Beyond total data privacy, this also means we have complete control over performance and quality—something that completely sets us apart from the competition. Lastly, most of our enterprise customers told us there was no way in hell their compliance teams would have approved sending sensitive customer data to OpenAI or a third-party provider. That’s yet another great reminder that doing really hard things for the right reasons pays off.

Pauliina: You have indeed managed to convince some bigger clients too in terms of these aforementioned considerations, and have grown rapidly in just one year, already securing six-digit enterprise contracts. Can you quickly walk us through how your system sets you apart from other (incumbent) customer feedback tools on the market?

Nima: A lot of people were surprised by how quickly incumbents adopted generative AI and launched AI assistants that could analyze customer feedback. While it might look good in press releases to launch quickly, these AI assistants are essentially compromises that need to live side-by-side with their legacy solutions without disrupting their core business. But AI assistants can only take incumbents so far.

AI assistants are compromises, and their customers start noticing this when they experience the quality of the AI solutions. I’m not exaggerating when I say that we speak with a logo customer of an incumbent every week, and every week, they tell us the same thing: the quality of the incumbents' AI solutions is really, really bad.

What I believe sets us apart from every incumbent and puts us light years ahead of the industry is the fact that we did the hard things first. We reinvented the experience from the ground up, built our own AI models, and developed a strong and unique methodology on how to manage customer feedback at scale. When you ask companies to automate work previously done by humans with AI, they have to trust that you can deliver on quality. AI analysis is one of those ideas that seems obvious and easy to implement, but in reality, it’s incredibly hard to reach human-level accuracy across different sources at scale.

Magic Feedback’s unique approach

Caroline Gattner (Maki.vc), Francisco Arias (Magic Feedback), Nima Vali Rajabi (Magic Feedback) and Pauliina Martikainen-Rahnu (Maki.vc)

Pauliina: One of your most recent standout features that I’m personally extremely excited about is the ability of AI surveys to ask follow-up questions in real time based on initial responses. Can you share more about how this works under the hood and how it helps unlock deeper customer insights?

Nima: Yes, when we look at the evolution of online surveys over the last few decades, we basically went from ugly surveys to pretty surveys, but there haven’t been any radical innovations. Now, with systems that understand the unstructured data that goes into them, we can also reinvent how online surveys work.

Our Magic AI surveys offer a radically better way to conduct online surveys. Our AI surveys understand both the questions you are asking customers and their responses, and they can ask great follow-up questions in real time if necessary to get better answers. We are essentially replicating what a great user interviewer does in in-person interviews, but doing it across millions of surveys, helping our customers get 40% better and more actionable insights than they otherwise would.

Vision for the future

Pauliina: And what’s next for MagicFeedback? Are there any specific upcoming features or expansions you can share with us?

Nima: We recently launched AI Actions, which means we now cover AI-powered feedback collection, analysis, and actions in one intelligent platform. With AI Actions, companies can create automated actions based on the feedback they collect and the insights generated from the analysis.

We already have examples of our AI proactively sharing reports with relevant teams when critical issues arise, creating recommendations to improve business metrics, and even producing production-ready code for review.

Pauliina: Okay – finishing off with going back to a higher level. Broadly thinking, what do you see as the next big frontier for AI in customer feedback and experience management?

Nima: It’s inevitable that we are moving toward a future where companies can simply turn on AI, and it begins collecting, analyzing, and acting on customer feedback according to their business needs. Imagine being able to instruct the AI on your company's tone of voice, values, and business objectives, and then having it manage the customer experiences of millions of customers at scale—sending out surveys, responding to reviews, conducting reporting, and implementing solutions.

Even if there are no further advancements in generative AI, we have made enough progress to reach this point within the next few years. In many ways, many of our customers are already experiencing this future today.

---

Curious to learn more on the topic? Visit Magic Feedback or connect with Nima Vali Rajabi to learn more on how they are revolutionizing customer experience. Have thoughts brewing on vertical AI? Share them with Pauliina at pauliina@maki.vc – she’s all ears!