In this presentation, Eugene Murphy showcases the Indeemo platform capabilities with a specific focus on Generative AI.

Transcript :

Hi, my name is Eugene Murphy and I'm the founder and CEO of Indeemo. Today I'd like to give you a quick introduction to the Indeemo platform. I'd like to share some of our learnings around Generative AI, which we have been very heavily focused on for the last twelve months. I'd like to talk about the client context in terms of the customers that we support, what we believe is driving a lot of the research that's happening on the Indeemo platform at the moment. I will then jump into a keynote presentation to give you a number of slides basically and talk you through the Indeemo platform.

And then I'll jump into the dashboard and I'll give you a live demo focusing on all of the Indeemo capabilities, and finishing up with Generative AI with the goal of giving you a, hopefully a very quick immersion into the power of video research and the opportunities that are available through Generative AI, especially for qualitative researchers and anybody focused on UX product or market research.

Generative AI

So I'd like to talk about Generative AI today because it's undoubtedly a once in a generation transformation in terms of technology and in particular the research industry.

As soon as I saw ChatGPT and the fact that my teenage daughter was using it to write cover letters for her job applications, I pretty much knew that it was the end of trust in text in the qualitative researcher survey context. And this email best frames what I'm talking about here. So midway through 2023, I received this random email from some random panel company and they're asking us, did we have any specific new technology to tort this copy pasting from cheating respondents?

So the answer was pretty simple to me really, because the answer is video. So if we live in an era where ChatGPT can write Shakespearean quality poetry, then in the research industry you can no longer trust text as a source of raw data and if you're in the business of providing guidance and insights to clients and you're still using text to collect the information in terms of understanding your research participants, then it's time to move on and at Indeemo, what we firmly believe now is that video is undoubtedly the future of understanding.

Again, if you are providing research outputs to clients and they're investing based on your advice, then going forward we believe you have to source all of your raw data in video so that you can stand over the guidance that you give and provide another trail for the client to go back to real human realities founded in consumer truths.

And if we were to say that all qual was going to go to video twelve months ago, you'd have laughed because everybody knew historically that video was extremely time consuming to analyze. It's very rich, it's very deep, it's awesome. Researchers love the richness, but they hated the time and they hated the effort.

But actually all of that has gone away because the flip side of Generative AI is that it makes it very, very simple to quickly analyze video at scale. If you've got a moment, you can scan the QR code there on screen and you can go and look at those reviews on G2 crowd. But the long story short is that Generative AI now makes it possible for you to gather vast amounts of video, analyze them in a fraction of the time using Generative AI.

Now coming back to Generative AI, everybody's got a million different opinions about it. We're pretty clear here in Indeemo in terms of the impact of Generative AI. It's gonna be an absolute boon for qual. It's going to massively empower qual researchers. It moves you from summary to strategy and consistently across all of our clients, what they're telling us now is that they get deeper into the data faster and they're spending more time at the strategic end of analysis versus the start of the process where they're summarizing or taking notes or just trying to figure out what's happening in the data.

Thirdly, we believe that Generative AI is going to be 10x the reach of Qual. Already we're seeing clients move from 20 participant projects to 80, 100 participant projects.

Generative AI now makes it possible for you to engage with hundreds of participants and be able to analyze the data in a similar amount of time to what you would've done historically with tens of participants.

And finally the last thing for sure is that we're seeing that Generative AI is enabling clients to do always on video research at scale.

So one of the trends we'll talk about here through the presentation today is this need to be constantly connected to your customer and with Generative AI and the video infrastructure that we've been building at Indeemo for years, we now have multiple clients who are running video research constantly for months at a time where they're connecting with participants on a daily, weekly basis and capturing rich video insights to supercharge the speed at which they can provide answers to their business stakeholders.

Client Context

Now secondly, I want to move on to the client context and what we believe to be driving a lot of the research that takes place on the Indeemo platform currently. So there are a number of key trends

that are driving the need for always on video rich insights.

The first is digital transformation. So it absolutely exploded during COVID and it continues to accelerate and essentially what this means is that every business that's going to be created from now on is going to be partially, if not entirely digital in nature.

What that means is that every customer journey is certainly going to be omnichannel. The second thing coming true is that you've got Gen Z and Gen Alpha now who are digital by default demographics and who demand seamless personalized experiences in every context. And that's driving a lot of the need for UX research and design and the creation of seamless omnichannel, digital and real world experiences.

And lastly, and certainly in the last 12 months, that's really accelerating the treadmill of understanding is this concept of exponential change. Again, it's being sent into warp speed by Generative AI.

And what this means basically is that, you know, change is happening increasingly faster and it's becoming impossible for brands to keep up. And in order for them to do that, they need a new way of doing research. They need a new way to understand their customers, and we believe that new way is video research always on at scale, which segues me on to Indeemo North Star.

So we believe that Indeemo that brands need rich real contextual insights and they need them faster and continuously. And if you're wondering what that looks like, I'd like to show you a video now from Winnebago which absolutely typifies this new type of research which businesses and brands need in order to stay relevant and innovative in a period of exponential change.

“The E RV two is without a doubt the most advanced RV we've ever built, the sustainable materials, smart tech integrations, intuitive design and connectivity and with something this groundbreaking, we knew we needed to explore another level of testing. So we created a pilot program that integrates interviews with a digital diary, not only to improve the design and tech, but to learn how people relate to an electric rv.

So we are going to make some dinner while we are waiting for the charge and how that differs from what we know about the traditional RV experience.”

We built a small fleet of prototypes and assembled a team of testers from coast to coast. It's unlike anything we've ever done before. We're giving the units to people and have them go off and camp as they normally would for a week, two weeks, and give us that feedback during the experience so we can catch them in the moment as what's going on, how they like the vehicle or what we might have to improve.

We're even sending key product marketing and leadership team members, myself included, out on the road to uncover insights that you simply can't get through traditional product testing. Input and reactions come in the form of screen recordings, video diary entries, snapshots and surveys, all captured on a digital platform in real time for our teams to analyze. We're now able to get much closer to the customer experience, actually seeing the world through their eyes and hearing about their experiences and their own voices.

“I love the aspect of having everything under one control panel.”

“One thing that would be helpful is that when you're plugging in at night, if you were to open up, if there was a little light that went on.”

“but super easy to flip back.”

We'll see things that they couldn't describe, it’s much more information than interviewing the tester after the fact.

We can study emotional concepts like range anxiety and see how they're different in say, Florida versus California, where charging stations exist in higher numbers. Through this new form of testing, we can truly understand the why and the how of each user's journey from their hopes and motivations to their successes and pain points while on the road and their reflections of their experience when they return home.

We've always been committed to listening to our customers and applying those insights to product development. The truths we uncover through this pilot program will impact the ERV two we bring to market the E-van life experience and every Winnebago RV that follows “

And just as context, all of the footage that you saw in that video was captured using the Indeemo app, which was given to the product testers when they went out piloting this new Winnebago ERV 2 product, which segues me onto Indeemo.

Indeemo Video Research Platform

So Indeemo is an AI part In The Moment Video Research platform used by MR,UX,CX teams to understand people, understand products, understand experiences, always in the context of everyday life.

You can use Indeemo in a number of different ways. So the first way really is what we call in the moment video research. And in this situation, what you're doing is you're leveraging our smartphone Instagram style app to connect with research participants in any context. And by sharing photos, videos, screen recordings or text, it allows research participants to bring you on their journeys to bring you into their lives, to bring you into their worlds and give you a deeper understanding of what's happening in their everyday lives.

The second use case for Indeemo is video analysis. So with Indeemo you can import videos in the form of interviews, focus groups, usability tests, and you can import all of that video into Indeemo secure research repository, transcribe it in 27 different languages, and then analyze the speed using Generative AI.

The third use case with Indeemo is mixed matter research where you can combine the ability to capture context rich insights using our smartphone applications. Additionally, you can merge in your one-to-one interviews or your focus groups that you record on Zoom or teams or in person and analyze both the mobile uploads and the recording interviews or groups as one body of data using Generative AI prompts.

On this screen, you can see a list of the various methodologies that our clients use us for. Now Indeemo uses across every industry and sector, from freelancers to small agencies to global consultancies to brands to governments. Excitingly, however, what we're seeing as I mentioned earlier is the power of Generative AI to enable always on video research. If you scan the QR code here on screen, you'll get to see a video from Sony Music where they talk about how they use always on video research for audience connection. And you'll hear James Bartlet tell an amazing story about how they were running a panel during covid, they saw the emergence of TikTok and they saw it exploding and it gave them the ability to react really, really quickly.

So how does Indeemo work? So what you're looking at here is a self-service interface for the Indeemo platform where you as a researcher log in and you can create projects in a matter of minutes. And in each project you can have one or more target groups. The target groups can each be given individual tasks or they can all be given the same tasks. You can embed videos and tasks, you can embed visual stimulus and tasks. You can also invite your colleague researchers, recruiters, and client observers into the project so that you can collaborate with them.

With respect to tasking, we support a number of different tasking scenarios. One, the most simple is what we call all at once tasking where tasks can be programmed to appear all at the same time and respondents can do them at any sequence at any time.

If you're doing longitudinal research, you will use schedule tasking, which allows you to program the start time in the end time of individual tasks.

If you're doing any mission or concept testing type studies, sequential tasking is where respondents register, they see the first task and only when they complete that task do re reveal the next task.

You can recruit participants in a number of different ways on the Indeemo platform. First of all, you can work with third party recruiters who can recruit the participants for you.

Secondly, you can recruit from your CRM by creating unique signup URLs for each target group in your study.

Finally, those branded customized respondent signup pages can be converted into QR codes and you can place those QR codes in specific contexts where respondents will scan those codes and register for the project.

Finally, what you're looking at here is the new recruitment interface on Indeemo. So we will be launching in recruitment directly via Indeemo in March 2024 and our goal is to enable you to create a project with self-service in less than an hour to quickly follow up a screener and then basically recruit your participants in 24 hours and be able to launch them into your study on the next day.

When respondents register on Indeemo, you have the option to create branded signup pages if you wish. And we also have a capability that enables you to upload your research consent forms, and respondents cannot register on the projects unless they double opt in via Indeemo app to take part without the need to have them sign any pieces of paper or go through any sort of a DocuSign process.

What you're looking at now is the Indeemo mobile app. Respondents say it's a bit like Instagram for qual that allows participant plea activities in any context and answers your tasks using videos, images, screen recordings, or text.

It's incredibly robust at capturing a lot of video on mobile. The video duration compression and bit race can be customized on a project by project basis. So if you really want to immerse yourself in the everyday context of your participants, the Indeemo app is a very powerful way to do it.

It's available in 28 different languages. One new capability that we've added recently is the ability to detect faces and blur them in sensitive research studies. So if you're doing anything in healthcare or B2B or colleague experience, for example, and you do not want to capture the faces of your participants, we have AI built into Indeemo camera and it completely anonymizes the research participants when they're taking part in your study screen.

Screen Recording is a really powerful way to understand omnichannel experiences. So what you're looking at here is a muted video of somebody looking at Instagram where they're searching for inspiration for recipes.

Screen recording is very powerful for usability testing, but also for increasingly omni-channel consumer journeys where you're trying to figure out how consumers interact with brands, the impact of influencers, and how they search for a particular product or service online.

As I mentioned earlier, you can also use Indeemo as a video analysis platform. So this enables you to upload pre-recorded video from your computer Zoom or teams and important to Indeemo where they're transcribed in 27 different languages and made available for analysis using Generative AI.

So the Indeemo dashboard has multiple rules and permissions built into it.

So the first one is a researcher where you can create projects and interact with participants and analyze data. We have a separate recruiter login where your recruiter can log in, add your respondents, and manage their activity on your behalf.

And finally, we have a client observer login, which is a read only login that enables your clients to log into Indeemo and really see their customer brought to life on a dashboard. The last thing about Indeemo is when you're using it for asynchronous research, you can engage privately, indirectly with each individual research participant using social networking style comments and push notifications.

Dashboard Demo

I'm now gonna transition to Indeemo dashboard and give you an interactive walkthrough of our platform's capabilities, including Generative AI. So I'm now logged into Indeemo research, repo and dashboard as a researcher.

What you're looking at here basically is all of the uploads from your participants for a click and collect curbside pickup project.

As you can see here in the project filter, however, it is possible to access multiple projects concurrently on the platform just by running the filter. And again, what Indeemo allows you to do is to collate all of the multimedia content from multiple research projects and store security in one research repo for you to access whenever you need to over time.

And just to give you some context on the research here, what we did was we had participants from three different countries go through a click and collect journey and document their experience all the way from ordering groceries online to collecting their groceries in store and reflecting on their experience and cooking the meal.

Now, as I said earlier, Indeemo is available in 28 different languages. So in this situation here, if I want to click on the French group, for example, all of the uploads from your French participants appear on screen. And if you click in here for example, and play this video, we're also transcribing the video in the native language of the research participants.

Now what you're looking at is all of the video selfie uploads from the research participants. And if you are using Indeemo for video surveys, this is typically what your dashboard will look like. Now, if I filter on a different task, for example, what's in the bag, I'll just do a quick search here. And now what I'm looking at is the board view for everybody's uploads, but without changing my filter, if I click on the collage view here, now what you're getting is all of the uploads that participants shared for the click and collect study where they're telling you what they purchased. And what this allows you to do basically is toggle between different uploads on the dashboards.

And this example here actually it's a video for the out of the box experience : “And so what I Liked about this include all of the things I wanted are…” You're standing in the participant's shoes and you're literally getting to see what they purchased and hear their feedback on their click and collect experience.

Again, without changing my filters now and still filtered on task six, if I click on the timeline view, what you're looking at now is a different way to see the same information, but this time it's organized in a reverse chronological order. Again, if I toggle here though from filtering on a task to filtering on a specific respondent, now what you're getting to see Mary's timeline of uploads from, if I scroll to the very bottom from her intro selfie to a screen recording of her searching for recipes on Instagram :

“Let's see what that's about. So this is probably, it might be a good idea to do the…” If I keep scrolling through here, now what we're looking at is her booking the collect experience: “check to, see what I have in, oh, book collection slot. I thought I did that already.”

So you can see here now she's a little bit confused around the user experience of the collection booking experience. She's going to the store, she didn't have a really good experience. Now she's in the car, she's sharing her feedback and you get to follow her diary moment by moment all the way through the cooking part of the journey and okay, you're right there in the kitchen all the way through to the awesome looking chicken.

Now, one of Indeemo’s unique capabilities is that when you enable ratings on a study, it automatically turns respondent uploads into an interactive multimedia journey map. So in this situation here, now what I'm looking at is I'm looking at Mary's journey rotate at degrees. And because we asked her to rate her feeling or her experience at every stage of her journey, what Indeemo automatically does is it turns Mary's uploads into an interactive journey map for you. And in this situation, we're just looking at Mary's journey map, but if I zoom out now, what it allows me to do is it allows me to see the individual journeys of every participant in the study.

So if I scroll down here moment by moment, again, you can follow the respondent's journey and see where any particular moment was positive or negative. But if you want to aggregate all of these experiences into one overall graph for all respondents, if I click here on the experience graph, what that does is it aggregates each individual upload from each individual participant in the study and creates a stage by stage aggregated experience graph, which gives you the highs and lows of all the participants working through the project.

One last thing to show you here is the ability for researchers to moderate and interact on projects. So as a researcher, if you have a question, you can type a comment in here such as how was your experience overall and hit return and what that does basically is sends a push notification straight to that participant, and if they reply at a later stage, then they, you'll get a desktop notification up here.

What that allows you to do basically is track all new uploads from all participants and any replies from respondents to comments or probes that you sent to them using push notifications. Now I'd like to show you how we're using different forms of AI to make it easy for you to understand video quickly. So the first example here is video transcription. So what I'm looking at here is screen recordings of participants shopping online. And if I click on the keywords drop down here, what we're doing here is we are using AI to extract text from video transcript and anything that the respondent types in and make that available to you for analysis.

So if I search here, for example, for chicken and do a quick search, what this pulls back is it pulls back a video transcript from one of the participants. And I can see that in the transcript button here the word chicken has been mentioned twice. And if I click here and scroll all the way down : “I want to order my groceries on”. Just pause there and scroll all the way down. If I click here straight away : “I know I have chicken I definitely need to get, and so I have my chicken, I have my…” So what that allows you to do basically is turn video transcripts into dynamic keyword filters and makes it easy for you to find specific words quickly in large amounts of video.

The second thing I want to show you with respect to using AI is our use of optical character recognition or OCR. Now what we're doing here is we're using computer vision to automatically detect and extract text from the actual body of the video. So if I type in something here again like chicken and do a search, now what it's doing is pulling me back to videos and it tells me here that the words chicken or chicken filets appeared on screen in this video for up to 49 seconds. If I quickly wanna see where those appear in the video, I click on the OCR button.

What we are doing is we are plotting any text that we find on screen along the timeline of the video, but because I've searched for chicken and chicken filets, what I'm gonna do now very quickly is I'm gonna scroll down and you can see all the dots move to the right.

If I click here on chicken : “see what” what you'll see here now is that it's actually appearing in the search filter and that's why you're seeing it so frequently along the timeline of video sugar.

So every time she searches, she's given the option for chicken. But if I wanna go down to chicken filets here, I'll keep scrolling. And you can see here it appears at two minutes and 38 seconds.

What we're doing now here, if I really zoom in on this, you can see the word chicken filets appears here on the packaging. So again, we're using computer vision to detect this text on the packaging in the screen recording of somebody shopping online, but we're making that text available to you so that you can search what's going on in somebody's screen recording.

So we see huge applications here for purchase journeys or usability testing. To give you one example, if you are doing a path to purchase and you are looking for a specific brand, we will pick that up using optical character recognition and we'll make it available to you for searching here. And you can search very, very quickly using optical character recognition. It takes you straight into that part of the video where the specific string has been seen using computer vision.

As I mentioned earlier in the presentation, it is now possible to use Indeemo as a video analysis platform. And in this use case, you have the ability to import any pre-recorded video from your computer cloud to cloud from Microsoft teams or cloud to cloud directly from Zoom.

Now here's a video of a focus group that we previously imported. As you can see here on Indeemo, now, when you import your multi-speaker videos, we automatically detect the speakers and we allow you to both name the speaker and give them the role of researcher or respondents.

If you do a lot of focus groups or IDIs or even usability tests, this is an incredibly powerful feature and it integrates seamlessly with Generative AI prompts, which I'll show you shortly.

Generative AI Analysis & Prompting

Okay, so let's talk about Generative AI. So this is the new AI interface on Indeemo. And what you're looking at here basically is that focus group I just imported. What we're gonna do is we're gonna write some prompts that show you the capabilities of Generative AI and the Indeemo platform.

With respect to Generative AI, what we have done is we've integrated with a secure private version of GPT-4 sitting on Microsoft Azure, where there is no training of the model based on your data. You own the prompts 100%, and you own the results of the prompts 100%.So it's an enterprise friendly version of Generative AI that is secure and private for your own commercial use.

What I'm gonna do now for the next few minutes, I'm gonna take you through some sample prompts and show you the results live so that we can show you the power of Generative AI when it comes to analyzing your research contents.

So on the interface here, basically I've selected this focus group, and now what I'm gonna do is I'm gonna write a simple prompt and just show you the user experience on Indeemo and what the AI returns in terms of results. So the simplest prompt you can write is summarize this transcript. So what I quickly do is I click submit, I choose the transcript, and then it sends it off security to Microsoft Azure for analysis using Generative AI.

So you can see the result of the prompt here. Now this is a one hour minute video and the result came back in about 30 seconds. So you can see that the AI generates a really quick summary telling you what's happening in the focus group. Now, if you wanna get more specific with your focus groups or interviews, the following prompt allows you basically to use the Generative AI to identify all the primary topics raised by the researcher and create a timestamped list of results for you to further interrogate with prompts.

So in this prompt, I've asked that to analyze the transcript and list the main topics raised by Ian. I've also asked that to include a timestamp for each topic. And while the prompt is running here, what we're doing now is we're leveraging the speaker identification that we've already assigned to the transcript on Indeemo and that sends Ian's sections of the transcript to Generative AI and it should bring back a really short list of all the questions raised by Ian, which essentially gives you a table of contents for the discussion guide of this focus group.

And you see that it's brought that back there in real time in less than seconds. And essentially it's giving you a section by section breakout of the focus group, which is really, really powerful.

So again, if I want to go into the first topic here, you can see that Ian has raised this. So if I click on : “so look, just to kick things off on our going through this, this focus group, I want to just introduce…” So now what it's done is basically gone straight into the first topic. It takes you from prompt to proof in a click using this proprietary timestamp technology.

Again, if we wanna scroll down to the four topic : “So for the four topic, I'd like us just to kind of discuss a bit…” It takes you straight in. Now that's a really, really powerful way for you to carve up your focus groups into discrete sections which you can analyze in a more focused way using the following prompt.

Now, in this particular prompt, I want to zoom in on the icebreaker section and I want to understand what each respondent said in response to Ian's question. So it can compose a prompt that tells it to start at the icebreaker section, which is mentioned here, and end at the first topic and analyze just this section of the transcript for me. I'm also giving you the specific instruction to create a table of results and telling it that I wanted two column table. So the first column is the respondent, and the second column is a summary of what each respondent said. So I'm gonna click on submit and send this to Generative AI. And again, in about seconds, obviously I've clipped this video, it's giving me back a quick tabulated set of results summarizing what each of the respondents said when Ian was asking them the icebreaker question. As you can see in the results Maeve mentioned that she'd like a meal of rib eye steak with mashed potato and gravy. So if you want to get that quote and extract that clip to put it in your reports, you'd write a prompt that looks something like this. So starting at icebreaker and ending at first topic, extract a quote from Maeve where she mentions rib eye steak.

And we're asking it also to include a timestamp for that quote. So it can go straight to that section of the focus group video. I click on submit and I submit it for analysis. As you can see here now the prompt came back literally in less than seconds and it's give me a timestamp where Maeve says she'd probably go with a full on rib eye steak. I click on : “And for the meal I would probably go with like a full on like ribeye steak…” It takes me straight to that section of the video transcript.

Now, if I do want to extract the clip from my reports, I click on the trim video here. I drag the needles up to this point of the video. UX is the same as doing something on your iPhone, for example. I'd literally click on download clip and I would actually extract that quote from Maeve when she's talking about ribeye steak and would allow me basically to put that into my reports and share that with my clients and my stakeholders.

Now to finish up, I'm gonna show you a couple of quick fire prompts just to give you an idea of what's possible using Generative AI on Indeemo. So in this situation here, I'm selecting four videos where the click and collect respondents were talking about their collection experience. I type in a prompt, it looks something like this. I submit the prompt to Generative AI, and in less than about 30 seconds, it's returned the table of results. It looks something like this.

Now, the powerful thing about our in integration with AI is that by asking it to bring back timestamps, it allows me to go from prompt to proof in the click. So if I wanna check the negative experience of Lucy, for example : “And there waiting with my boot open and it was raining, so it wasn't ideal…”

I go to Adam : “All it took, There was 20 minutes waiting time from when we arrived and I said the first half…”

If I go down here : “The Collection times that I could select that, I just found it quite annoying that I had to wait until the following day.“

So I hope that's giving you a quick introduction into the Indeemo platform and the power of Generative AI to do video research at scale. As you can imagine, it's very difficult to squeeze everything into such a short demo. So if you have any questions, looking forward to answering them now with Mike.