Knit Virtual Session: State of Innovation in '24: The 4 Biggest Areas AI Will Help the Innovation Industry Innovate this Year

How AI is Helping Innovate the Innovation Industry

Editor’s note: This article is an automated speech-to-text transcription, edited lightly for clarity.  

On February 1, 2024, The Quirk’s Virtual – Innovation event kicked off. The event was a series of virtual sessions presented by researchers from all over.  

Aneesh Dhawan, CEO of Knit, started the event off with a presentation on the state of innovation and how AI will help the industry to innovate this year.  

Watch the recording or read the transcript below.  

Session transcript 

Joe Rydholm: 

Hi everybody, I'm Quirk’s Editor, Joe Rydholm. Thanks for joining us today and welcome to our session “State of Innovation in 24: The four biggest areas AI will help the innovation industry innovate this year.”  

Just a quick reminder before we get started that you can use the chat tab if you'd like to interact with other attendees during today's discussion, and you can use the Q&A tab to submit questions to our presenter and we'll address them during the Q&A portion afterwards.  

Our session today is presented by Knit. Aneesh, take it away.  

Aneesh Dhawan: 

Awesome. Thank you, Joe and thank you all for attending this morning. We're excited to be kicking off the virtual event today.  

My name is Aneesh. I'm one of the co-founders and the CEO of Knit, and we're really excited to be talking to you all today about the four biggest areas AI will help the innovation industry innovate.  

For those of you who are hearing about Knit for the first time or are less familiar with who we are, think about us as your AI research assistant. We can help you combine the power of quantitative data, video data and AI to help get a quicker and more holistic understanding of your target audience.

So again, today we'll be talking about four of the biggest areas where we see applications for AI to help the innovation research process. But before we dive in, we want to just start off with a quick overview of where we believe the innovation research process is today as we head into 2024.  

Research and insights for innovation is obviously critical. It helps us understand consumers as people, helps us pinpoint and forecast consumer trends for our business. But one thing we're seeing across a lot of our innovation research on customers is an immense amount of pressure, both internal and external, for their businesses to innovate.  

The external pressure is coming from consumers who are changing their behaviors rapidly in this digital age, the internal pressure is coming primarily from stakeholders who want to make sure that their businesses are keeping up to speed with these changing consumer preferences.  

And that's why as we head into 2024, we believe that the theme for this year will really be around speed and agility. Agility, making it more important now than ever before to have speed and agility in how innovation researchers are running research. And that's why we're so excited and so bullish on what AI can unlock for this industry.  

By automating the grunt work and research, we believe that AI can help condense a research process that typically took weeks or months in as little as a couple hours or days, and that's rarely where we're seeing some of the biggest value being driven with AI in the innovation research process and bringing some of those meaningful insights to stakeholders in an agile way.  

Ultimately, this helps shorten timelines across the board, helping you get products out to market faster and staying ahead of your competition. You can hear how AI and AI tools have impacted some of our researchers on the platform today.  

Now, outside of speed to insight, one of the big opportunities and perhaps an even bigger opportunity we see with AI is AI's ability to empower researchers to help raise the bar by running better research. And by better research we mean more rigorous research, more creative research.  

We'll touch on a lot of the opportunities that AI enables today in our conversation as well. But needless to say, we are incredibly excited about all the different applications for AI across the innovation lifecycle.  

From wide space exploration to packaging testing to consumer journey mapping, we'll cover a lot of how you can run those types of research and where AI will impact those types of use cases in our conversation today as well.

With all that said, let's go ahead and dive right into today's topic, which is the four biggest areas where we believe AI will help the innovation industry innovate in 24. What we'll do over the next couple minutes here is I'll walk through some of the areas that we, as an AI platform, are most excited about when it comes to applications for AI.  

We'll walk through how we've brought that to life on the Knit platform, and then we'll actually walk through what those workflows look like today. You can see how other innovation research teams are actually leveraging these AI tools to make their workflows more efficient or to help raise the bar in the research that they're doing today.  

So the four big areas we see where AI will help innovators innovate in this new year are really around three areas in the research process. The first is the study design process, the impact there really being around how can we help researchers ask the right questions in a survey.  

Then the analysis process with the areas we see the biggest impact being around making sense of unstructured data. How do you take thousands and thousands of open-ended text responses or hundreds of hours of video responses and make sense of all that unstructured data, as well as summarizing the key insights. Once you can make sense of that unstructured data and actually digest it, how do you actually pull out those key insights and those key takeaways for your business?  

And then finally, in reporting, how do you actually act on those insights in a more effective manner or deliver those insights to your stakeholders in a better or faster way?  

We'll start today off by looking at the design phase. When it comes to study design, we think there's a lot of compelling use cases for AI. The philosophy behind how we built our platform when it came to our AI survey generator is really about how can AI collaborate with the researcher to generate a survey design or a study design that's not only faster, but that is also more creative in exploring different question types or different approaches to the question.  

And so how can AI really collaborate with the researcher and be a true partner rather than the researcher and the AI working without each other's collaboration?  

When it comes to study design, the way we've built out our platform to do that starts off with giving the AI the context that it needs. So very similar to how you would write a research brief, you can share with the AI some of the key aspects of your study, your business overview, your research overview, your research objectives, as well as your audience criteria. 

The AI then takes a first stab at generating the draft, and that's where things get really interesting.  

We've trained our AI to be creative in how it's coming up with these surveys, looking at over a dozen different question types and generating questions across those question types as well as just making this incredibly customized for our brands. So, generating it in a Microsoft Word document like tool where the researcher can go in and customize this to their liking.  

Now, at the end of the day, what this is, is a really collaborative process with the researcher able to continue to prompt the AI to come up with different approaches to those questions or rewrite questions through different question types. It's a back-and-forth process here with the AI and the researcher collaborating very similarly to how a researcher would collaborate with someone on their team.  

Now the efficiencies here continue on actually programming that study.  

This AI survey generator can then program the survey that's been designed directly into a survey tool. Again, this helps shave off those hours that you would spend taking a survey from a Microsoft Word document and implementing it into a survey tool. 

The survey design process is a great example of how AI can actually help not only speed up the process, taking something that might've taken at least a day and doing it in as little as a couple hours, but also make that process better. Really helping you work with the AI to think outside the box, unlock different question types or different approaches to questions that either you don't have the time to do or that you might not be able to do because of existing biases or blind spots.  

The second aspect of this process is analysis. And I think this is where the meat of AI's applications in the research space lies. Within analysis, we see two really big opportunities. 

The first is around making sense of that unstructured data. How do you take thousands of open ends or dozens of hours of video and structure it so that people can easily read and digest those key insights?  

And then the second opportunity is pulling out those key insights. How do you turn that kind of raw data that's now structured data and actually pull out those key takeaways that are meaningful for your business and meaningful for your business objectives?  

There's a handful of different opportunities and different ways that we've brought that to life on the NIP platform that I'd love to walk through in the next couple minutes. 

The first is around how do you make that large amount of unstructured data meaningful? And so, one area that we've brought to life is in our AI generated insight summary. 

What you're seeing here is this summary that comes from analyzing, with the AI reading transcripts of hundreds of videos in this case and generating a summary that's about a paragraph or two paragraphs long.  

And the benefit of this is that you no longer have to go through and read through all the raw data, whether it's watching all those videos, reading all those open ends, you have that 30,000 foot overview at your fingertips through the AI.  

Now, not only does this make the process faster, but it actually is more rigorous than the traditional method of analyzing these large data sets. And that rigor comes in the ability of actually citing all these sources and showing exactly where that data and those themes are being pulled from.  

What you're seeing in this screenshot is across this summary there are little green boxes and those are the citations. And the citations tell you the frequency of those insights.  

How important is that insight to your overall objective? Is this something that's been brought up by 60 people or is this something that's been brought up by five people? And if you want to double tap into that, you can actually see where it's being pulled from, what specific videos, what portion of that transcript is it actually pulling that insight from to generate these themes.  

Now to double tap on that, if you want to dive a little bit deeper into those themes and sub-themes, one way we've been able to do that is through our AI contextual analysis tool, and that's another really exciting application for AI in the research space. We've been really surprised by just how well AI can mimic what a human researcher does when it comes to coding unstructured data.

Now, I don't think that AI will ever be able to fully replace how a researcher actually analyzes unstructured data, but the goal here is if it can do 70 to 80% of the grunt work there, then the researcher frees up a lot of their time to not only tweak it and make it work within the context of their business, but actually spending time where it matters most, which is bridging the gap between insights and action. 

Our AI contextual analysis tool is kind of our AI coder. It's our AI coder that's able to go through all this data and code it like a human would code it. And what you're seeing here is the ability to code all that unstructured data here as the key deems and then the ability to kind of drill down into that into some of those sub themes. You can see the frequency of all those themes, and when you click into the themes, it's a very similar philosophy to the summary where you can actually see exactly where all that data's being pulled from, what specific videos, what specific aspects of the videos is it pulling all those insights from.  

And again, this allows you to really collaborate with that AI and be able to customize it to your liking, make sure that the output of the AI fits within the context of your business and get it from that 70-80% output to a truly 100% output that works for your business and your research objectives here.  

Now, what we saw across a lot of these examples was AI's ability to quantify the qualitative data. And what's really exciting about that is because AI can quantify qualitative data so quickly, it can read both quantitative and qualitative data as one larger data set, and this gives you the ability to, we like to say “Knit” together. Quant and qual really at the end of the day kind of cut the qual by the quants and the quant by the qual. 

And what this unlocks for research, what we've seen this unlock for a lot of our research partners is the ability to get a more holistic overview of the customer, not only looking at the customer through the lens of a quantitative lens or a qualitative lens, but actually looking at them holistically through both a quantitative and a qualitative lens.  

The last application for AI we see within analysis is one that we're probably all very familiar with. If you've ever had a chance to play around with ChatGPT or any of those other AI tools, and that's changing the interface or the way that researchers interface with data, traditionally, we're all looking at data through the lens of charts and graphs, but with AI, you're able to change that interface into a more natural interface, for example, a chatbot. 

The way we brought that to life on the Knit platform is through our AI chat bot called Aida. And the great thing about Aida is you can ask Aida very pointed questions to get to that underlying insight you're looking for in your research. And this helps you accelerate the speed to insight. You no longer have to go through the traditional workflow to get to that insight. 

You can kind of fast track that and go directly to the AI interface directly with the AI to get to that insight. And one of the great benefits of being able to analyze both quant and qual data in one dataset is the ability to chat with the AI and ask those pointed questions with both the qualitative dataset but also the quantitative dataset. You can ask specific pointed questions around the quantitative data, around the qualitative data or a combination of the two. 

So across all of these tools and all of these features, the idea behind what AI can unlock really comes down to that speed and that raising the bar that we were talking about earlier today. What we found in our research was that traditionally, about one hour of video data was taking 10 hours to analyze before our AI tools. Today, that same one hour of video data is taking just 10 minutes to analyze. It's making the process much more efficient and incredibly faster.  

The second aspect of this is it's making the research better. It's bringing more transparency and thoroughness and how qualitative themes, for example, are being coded as we talked about with the citation and summary tools that we just covered. But it also unlocks a different way of thinking about how we run research because now you can go a lot deeper with your customers without having to worry about analyzing all that traditionally qualitative or unstructured data on the backend. 

A very basic example of that is what does this unlock for you in your business and in the research, you're doing when you can analyze five open-ended text questions as quickly as one open-ended text question. There's a lot of really great opportunities there to start digging a lot deeper with your customers when that field has been leveled there.  

And lastly, the big opportunity or the final opportunity we see is really around the reporting when it comes to insights, and we see the applications for AI and reporting kind of in two major buckets.  

The first bucket here is around more quality-of-life improvements. For all of us that have put together a research presentation, there's a lot of grunt work that goes into it, building out the graphs, designing it, making sure it fits the branding of your organization. A lot of that can be automated with AI.  

There's plenty of incredible AI presentation tools out there. If you haven't taken a look at them, I would suggest Googling and finding some. But it really helps take a lot of that grunt work off your plate, and AI can generate a lot of those, get the polish that you're looking for in those presentations.  

For example, with the Knit platform, we've built a lot of that into our platform where the AI can generate those graphs for you, make a lot of those kind of branding and design elements as well.  

The second key bucket there in reporting is around actually generating insights. So, AI's ability to read through the data and generate insights for you based on your business objective or research objective. 

A great example of that is something we built into our tool where we're actually able to read through a graph and generate a two to three sentence insights for you.  

Again, the idea here is not to replace the analysis that you would be doing, but to actually speed up that process to get to that final presentation, that final report for your stakeholder. Imagine being able to just see a two-to-three-line top line summary of all the different graphs that you've generated on the platform. It allows you to just build that narrative in your head a lot faster and actually end up building that presentation that you might to want share with a stakeholder a lot faster as well.  

Now we see these efficiencies in both the quantitative data as well as the qualitative data. So similarly, there's a lot of quality of life improvements that AI can make when it comes to sharing out qualitative data, whether it's editing these kinds of sizzle reels and show reels, fixing AV quality, making sure that the captions match what someone is saying and the transcript is solid and accurate. 

A lot of that can be solved through AI tools, many of which we've actually implemented within the Knit platform here, but also being able to help you generate those insights. AI's ability to identify these are the key videos that you would want to share with a stakeholder based on what you've shared with us as your business and research objective.  

At the end of the day, reporting can be one of the longest in terms of the time intensive process when it comes to running research.  

Again, traditionally we were seeing that that process could take anywhere from three to five days on a good day. What we're seeing and hearing from our customers now is that the reporting process is being done in as little as 24 hours. But the best part here is what it unlocks for the researcher.  

You now have so much more time to focus on ‘how do we take these insights and actually bridge insights to action within our organization rather than having to worry about some of that grunt work that goes into building out these presentations and reports for your internal stakeholders.’ 

So, just to wrap this up and put some numbers on it. Before releasing our AI tools, which we launched last year, we were seeing that the average project on Knit, it was taking about four weeks from start to finish. Today, the average project on our platform is about 48 to 72 hours, and we're hearing this as one of the biggest value drivers for our partners that are leveraging our AI tools. 

For example, one of our partners shared that the typical qualitative project they were running off our platform was taking them about 300 hours, about two months of one full-time person's work today that's being done in about 10 hours, about a day and a half. 

But to just drive this point home outside of speed to insight, I think the biggest opportunity here is really about how this allows us to run research differently, how it allows us to level up the types of research that we are doing because of what this technology unlocks for us. 

And you can hear that sentiment from some of our partners here as well. As we head into 2024, they're really excited about what AI can do for them in terms of delivering more well-rounded reports and delivering more quantity and quality in their research. Again, using that example of the open ends, right?

If you can ask more in depth questions, if you can dig deeper with your questions in a survey because AI can help you analyze it better on the backend, then you don't have to necessarily make as much of that tradeoff between depth and speed that we've historically had to make in the research industry.  

As we wrap things up, I just wanted to share today, we talked about some of the biggest impacts we see of AI in the market research space. We are truly an AI native end-to-end platform. There's a bunch of other areas where we've applied AI that have delivered great results and value for our customers. 

One that we didn't talk about today was how we've used AI to vet data quality and sample quality. So that's always something we're happy to have a conversation around and maybe we'll share at a future event, but I'll kick it over for Q&A.  

If you have any questions, feel free to reach out to us or check out our website. We really appreciate you guys joining us here for the first presentation of The Quirk’s Virtual – Innovation event and looking forward to answering any questions that we have. Thank you.