A masterclass on designing intelligence flows for the AI-enabled enterprise
Hi, everyone, and thank you so much for joining us. I am Maria from the marketing team here at Market Logic, and I’m joining you today from Berlin. We have folks from over 30 countries joining us today, and most of you are in The US, UK, Germany, Sweden, the rest of Europe and Hong Kong. So good morning, good afternoon and evening wherever you are. Just a bit of housekeeping before we jump in. We have about forty five-fifty minutes together. The session is being recorded and we’ll send the recording and the slides to everyone who registered. If a question pops up as we go, just drop it in the Q and A box anytime and we’ll take a look at them at the very end and answer them. We also have a couple of quick polls along the way, please take a second to answer them. Today, it’s all about how leading enterprises are wiring market intelligence into their AI. And I cannot think of two better people to walk us through that. So without further ado, I’ll hand it over to Olaf Lenzmann, Market Logic’s co founder and Chief Innovation and Product Officer and Nick Papakonstantinou, Senior Product Manager for our AI and Innovation portfolio. Olaf and Nick, over to you. Excellent. Thank you very much, Maria. And welcome to the session, this master class, but maybe in more simple and straightforward terms. We want to talk about in this age of AI and AI permeating more and more of the enterprise. How can we make sure that insights not only continue to be relevant, but how can we maybe use this as an opportunity to make sure insights actually are amplified and are anchored in this new age of working with AI across the enterprise. And how can technology help their MCP is the magic word here. We’ll talk a little bit about what that is. And hopefully we will leave you with some inspiration on how you can use technology to really use this moment as your companies, your enterprises develop their stance and their architecture of AI find themselves, how you can use this moment to make sure insights are firmly embedded and become part of the fabric. I’m very happy to be here today with Nick together. We’ll take you through this webinar. I will start with a brief intro, contextualizing again where we come from, what MCP is, and what the general structural approach is there. And then we will talk about four specific use cases, component use cases, if you will, that look at specific capabilities that you can integrate and how they then hopefully move the needle at the bigger picture. So let’s start by taking one step first back and from our perspective at Market Logic with our DeepSights product portfolio, what do we aim to achieve? What is our mission? It’s first and foremost helping our customers to connect all the knowledge and insights they have to govern it into one source of truth, And then to take that and embedded and activated in the organization, which is what we’ll talk about today. And lastly, to learn how and what the demand is in the organization to help prove what the ROI and what the contribution of insights is. But as I said, what we want to focus on today obviously is squarely in the middle. It’s embedding and activating those insights. And previously, if you will, embedding and activating meant bringing them to human stakeholders to bring the insights basically towards the decisions, the good old insights democratization. And now with AI entering the picture, we have situations where AI is an intermediary between us with the insights and the stakeholders or where even the AI is the ultimate consumer if you do have agentic processes that maybe start doing innovative explorations, etcetera, etcetera. So how can we, in this world, bring the insights to the decisions? And first of all, why is that important? Now it is important obviously because AI, when it works on a problem and attempts to help or even decide in some cases maybe, it needs context. It needs to understand what is the situation. And of course, there’s a lot of context that it needs that is just rep… That is representing your company’s enterprise reality, internal context reflected in documents and emails and other systems of record like CRMs and business intelligence systems, etcetera, etcetera. And typically, the copilots of the world are already connected to that and have a good view into that. And when we talk about copilot today, take this just as one example, which may as well be your Gemini or Claude or Amazon Spark or other assistance system that that your company deploys. So Copilot, as an example, has that view on enterprise reality, but what it does not have immediately is a view on the external reality. What’s the market, the consumers, the brands, the competition looking like? And it needs this rounded picture to have the really the, what we call the golden context, the full three sixty view, so it can give the best advice and can support decisions in the best possible way. And specifically when it comes to insights, we also see from some sources that a majority of stakeholders expect AI agents to start mediating access and use of insights in the future. So it’s really important to find our way with insights into that world. And what does it mean? We think it means that this market intelligence, the insights, the knowledge they belong as a first class citizen into the AI stack as appear to other sources and other contributors. And of course, there are different sources, different remits. There is the whole world of what your… What customers, direct customers of the enterprise do, which typically resides in customer data platforms and the like, which is an established and existing source. There’s a lot of business data and BI information that is an existing source, but there is also exactly this why and how markets and consumers behave, and that is what we with DeepSights would provide, and that also needs to become this peer source of insights. So that’s the context. And now if we think about how can we bring this market and consumer knowledge into the AI world, we see for prototypical components, basic use cases, if you will, as we work that we work on with our customers, and that we want to bring to life for you today. The first one is, let me say straightforward, which is plugging in the insights to the existing enterprise AI assistant ecosystem so that for example, users can go to their Copilot or their Gemini assistant or their Amazon Spark assistant and ask questions and get the same knowledge and insights that they would get if they were to come today to our platform. Once that’s in place, also an important use case and an important capability for many of our customers is to use this now as a funnel to bring together also other partners, other sources of information that also provide live data and live perspectives, but to bring them in in one coordinated and connected way and perspective, and we’ll show that also in a moment. Then thirdly, the question is how can we take a step further from waiting for a user, for a stakeholder to ask for insights and be more proactive and actively embed the insights in the work that’s happening with AI. And that’s what we’ll talk about with the revenue use case. And then lastly, we will also look at delegation, which is a first step when we go think about users, if you will, being actually agentic systems that, for example, design campaigns to other things, and then might delegate, say, certain work, for example, for verification back to insights agents like we provide with DeepSights. That’s what we wanna talk about today. But now as an intro, we have a short video that tries to break to life a little bit what is all this MCP about and how does it fundamentally work. AI assistants are incredibly smart, but they don’t automatically have access to the systems and knowledge your business relies on. Traditionally, connecting different AI assistants to different systems meant building and maintaining separate integrations. That’s where MCP comes in. Model context protocol. Think of it as a common language that lets compatible AI assistants connect to external tools and information. Now imagine asking your AI why sales are declining. It can check internal business data, then use MCP to ask DeepSights what is happening in the market. The request leaves the assistant and lands in DeepSights. DeepSights understands the question and checks your permissions. Using your company’s own semantic entities and knowledge graph, it finds the best matching primary research, syndicated reports, and news. It creates a synthesized response with watch outs and sends it back to the assistant. The assistant can use it on its own or combine it with the rest of its sources. MCP makes the connection possible. DeepSights brings the market intelligence And your AI assistant brings the information together while respecting your access permissions. All right. Yeah. I hope that gives a little bit of initial context how and why MCP as this magic acronym works. And just to reiterate, it means first and foremost, there is one way of connecting to a multitude of AIs that can then leverage what this, for example, DeepSights accessible through MCP makes possible. And so it can be used both in systems, it can be used internal agentic systems, etcetera, etcetera. And so in practice, of course, there is a little bit of setup that’s required. There’s a little bit of setup required with IT that we will be doing together with with your IT. But once that’s done and in place, it is available. It can be used in any assessment. It can be used by other agents. Importantly, however, also, it does respect, as we heard on the video, access permissions. Meaning, you… If user a would would be querying through their AI and and using this capability, they would exactly get the information that they are entitled to see. And also, as your AI, your Copilot, or Gemini would be free to decide when to make use of of DeepSights and the tools, we also provide those instructions to it to give it clear guidance when to leverage this kind of information, and we’ll come back to that in a moment later why that’s important and how it can be used to be very powerful. So there is no coding development needed, there is some setup needed, but once that’s done, it gives us, as we say here, a seat at the table, it gives us a way to amplify the insights into that ecosystem. And having said that, we now wanna dive into the first use case, which is about answering questions not on DeepSights, but rather in your infrastructure, and Nick is gonna take us through that. Great. Thanks, Olaf. Good morning. Good evening, everyone. I’m Nick Papa Constantino. I’m the senior product innovation manager at Market Logic Software. What we will discover next is the first case that we see, how we bring the insights discipline and copilot via our MCP when users ask, for insights. And let me start with the problem. The reality is your AIs respond to all your market questions, I would call a little not as intelligently, and they make it sound very confident. So honestly, I think this is a little wrong. They may surface out at its studies from very specific or many very dedicated type of, I would say, questions like what what’s driving Gen Z snacking in Germany. And we can see very easily that, for example, an… A source might be wrong. Uk study may be used for, generalizing, an insight finding for Germany, and, some of the sources could be widely outdated. Obviously, this is not out of the box something in business user will easily understand, but an insights professional will actually spot that very quickly. And just to zoom out a little bit more, Copilot as one example or any other agent is a very good agent. It can provide the documents back, but what we think it’s missing is what we call the insights discipline. So in many different ways, we can see again, as we show in the previous slide, the right study may be missed when finding the right document, maybe the wrong market is used. Syndicated reports are completely missed. Code names are not necessarily recognized from the agent unless you supply additional context from the, let’s say, data prompt layer. And latest might all not always mean that most recent document or insight that I need. So all of those problems that we see, they’re not necessarily access related problems, but they are what we call the market context that is completely missing. And this is basically where DeepSights comes in. And we do have our, let’s say, greatly improved market context layer that we started rolling out customer by customer already that greatly enhances all the responses that Copilot or relevant agent agents, utilize. This slide is a little complex, but it is a little complex on purpose. This is literally your team’s expertise made usable by AI. So nothing here or thing that we’ll be pointing out here will surprise the insights people in this call. It is what you already do in your head when a stakeholder asks a question like, is our digital and social content working? Which can be also questioned as is our tick… TikTok content doing anything? So what happens is before any type of search, DeepSights translates the query and provides a shorts… A short research brief. And this brief is provided the same way an insight or senior insights manager would frame a question. So it’s not necessarily only asking about is our digital and social content working, but is actually asking for more things. And what we basically do with this brief is we delegate back to DeepSights the, let’s say, the research flow that needs to follow that basically ask it to find out specifically all the different types of questions that need to be answered out of the query of the user. It basically points out where the evidence exists, which type of studies provide this type of evidence, and also connects it to the actual programs run-in your company. Again, underneath all of this sits all the expertise of your senior practitioners with all the relevant watch outs as an example in this specific case, because we’re talking about social media, likes and shares, and not necessarily execution metrics that do not have, let’s say, brand effect. Again, all of those things are part of the… What we call a research brief, that is basically provided to DeepSights to better search. Again, this is not the concept. This is, a market layer or maybe a context layer. We have built around hundreds of business questions we have encountered, and we have tested. So those are the results of the test. So we took a search with DeepSights and a search without DeepSights. And what we have found out is that a search with DeepSights looking at the same types of documents provide the best answer four times more frequently. And with best answer, basically, we mean that the relevant type of questions and the… Sorry, the relevant type of documents are already found within the evidence to be supporting the answers that we provide. And again, that already gives you a little bit of an idea on how much better it could be just to have an an intelligence layer on top of research to do more, I would say, informed decisions on what is a good evidence. So as we all understand, this is not necessarily an access problem. This is essentially the insights discipline completely missing. And what I would like to do next is I would like to give you a very quick demo using our own Copilot agent to see how easy it is to, work with, DeepSights connected. So allow me one second. Transition. Alright. So we have created this Copilot agent. You can also create your your own Copilot agents, or you can connect DeepSights in your own Copilot chat. But for this specific demo, this is just an FMCG type of, I would say, agent we created because it connects to our FMCG type of data in our system. And what I will do is I will ask a market question, which is… Let me have it here. What’s driving Gen Z snacking in Germany? And, essentially, what I would like you to do is understand that this is a market question and route all the responses, back to DeepSights. What DeepSights does, runs its semantic layer across the query, tries to understand what the research is about, and finds the relevant primary evidence. So all your primary research, your syndicated reports, your trusted news, and creates one final, I would say, response. So all the information used in the evidence that we provide on the answer is basically clearly cited on Copilot the same way we also do it on on DeepSights end. And we also supply in every single response we provide what we call a watch out. So essentially, what this is is an additional layer to give, let’s say, one more quick note on the quality of the sources before anything, is used by, an insights professional or a business user. Again, gives them a little bit more confidence on the response. If you would like to just revisit any of the information provided here, you can click… You can necessary… You can actually click on one of the, let’s say, dedicated citations here, and you will land directly on DeepSights just to validate where the evidence is actually coming from. And, again, this is just a very short video to show you how easy it is to just have Copilot or your agent connected with DeepSights to work seamlessly. So let’s move a little back now to to the presentation because there’s one final remark to… That I wanted to make. So this is basically gives you an idea on how many different types of agents we have connected DeepSights to. So we’re using MCP. We have connected to Copilot, to GPT, Cloud, also coding agents like Cloud Code, like Cursor, Gemini, Clean, also MCP gateways and custom agents. So we do have a really abundance of connections that we have made. Connecting to DeepSights, again, we supply all those agents with the… What we call the market context skill. So DeepSights will always be consulted before we provide any type of, let’s say, response back to the agent. So that pretty much covers the first part, which is the ask case. But I know already that we have here, let’s say, agencies that are participating in this webinar. So this is more just to give you a little bit more, let’s say, information on how we aggregate all the external signals into our own final response and how we make that, let’s say, responding more, let’s say, meaningful considering all the different signals that we get. So the example we will cover today, again, it’s a little more targeted to the latest two, I would say integrations we made with Synctasia for social listening and Next Atlas for forecasting. But it will give you an idea what is the challenge that we are facing, but at the same time, how meaningful it is to provide this type of, let’s say, intelligence into our DeepSights semantic layering and responses. And the business reality is that enterprises work with so many different tools. And they… Because they have so many different types of use cases. Imagine if you had for each one of those tools to use a different MCP, a different AI to get one response. The problem is you don’t know necessarily what a trust and then how everything comes together. So this is a really bit of a, I would say, hole in the hole, in the whole system. So this is where DeepSights comes in. So it considers all the primary data, assist your subscriptions, your news, your social listening, if you’re subscribing to social listening tools, your forecasting, and then aggregates all the insights, creates one final response, and provides it back to, let’s say, MCPs so they can be consumed by any type of agent. So it really creates the one answer you can actually trust. And with that being said, all I wanna do is also show you one example how that comes to life in DeepSights in a similar fashion. Just give me one second to transition again. Alright. So as we said before, we made, let’s say, we have… We are tapping using our MCP. We’re getting data over from, Syntasio and Next Atlas, Syntasio for social listening, and Next Atlas for Foresights. So basically, what we’re doing here is I’m asking a very specific type of question, which is targeting social listening and forecasting just to understand how everything comes together. So the question I will be asking is what are US consumers saying about plant based foods and where is the categories heading? US consumers and what they are saying is the social listening and the category heading essentially, it is gives us a little bit more of a understanding on we’re looking for forecasting type of information. So we get the final response back from DeepSights using Copilot. And we can see here that all the evidence coming from social listening is attributed to Synthesio. And all the forecasting is basically utilising the Next atlas type of types of, let’s say, insights. What we can do again, because again, we are citing all the evidence that we get from all the different partners. If you do have an SSO with them or authentication, essentially, what you can do is you can click on the necessary link and you can land to that specific evidence directly without any sort of, I would say, login path. So, again, this is the one aggregated layer that we’re offering on DeepSights sent to consider all the primary, all the syndicated, all the partner signals in a single response just to provide it back to any enterprise AI agent so they can be harnessed in any sort of a workflow or for the final response. And, well, with that being said, just to close this segment, there is basically a little bit of a call to action, I have to say here. Again, if you’re an agency that works with many different enterprises and you do have your own your own signals or your own important status, your own AI. What you can do is we can connect together, and then we can find avenues to bring in your data over to DeepSights using our MCPs. And then, of course, find the proper placement in our harness, attribute the sources back to your own source name, and then eventually offer it via our DeepSights connection to any enterprise around the… Our, let’s say, common ecosystem. All we have to do, again, you can send an email to partners@marketlogicsoftware.com, then we can have a discussion to see how everything will come to life and how jointly we can bring more meaningful evidence to the customers and the enterprises around the world. And I believe that basically closes my queue to give it back to Olaf. Yeah. Thank you, Nick. And I just wanna pause here for a second before we move on to the third third piece. Just to recap, I just want to emphasize again, what we saw in the very first session, in the first segment about asking questions, getting the answer in Copilot, and what Nick showed about this, as we called it, mapping of understanding how to get from a business question through what evidence is needed, answers need to be found to the individual studies that may be the reference sources of truth in your organization. That is really encoding the knowledge and the understanding of how to use insights, of course, together with the customer, with you. And that is the foundation why this gives so much more reliable, trustworthy, and following the insights, discipline, and governance answers within the AI ecosystem as opposed to just exposing all the raw documents to an AI and then hoping for the best in terms of how it would use it. And that is really the foundation for the use cases we saw. So having reiterated that, let me move on. And we just saw Nick asking questions, a copilot, again, could be any other assistant. But now let’s think about how can we maybe take this a step further. How can we go from being a reactive provider of insights and information that still sits there waiting for the stakeholder to come and inquire and ask for that information, how can we take it a step further and become proactive? And how can we maybe leverage this new AI age to achieve something we haven’t been able to do before? In other words, stepping back, looking at the problem, and thinking about what are all what is all the work being done in the enterprise today? There’s many, many, many, many different plans, big plans, small plans being done every day across the organization. Can be very strategic plans, they would typically be subject to some structure process that maybe has stage gates that maybe mandates use of insights, fine. But there’s also a lot of tactical and smaller planning and understanding activities that would need insights for a full picture, but that entirely rely on the person working on them thinking to look at insights. And if you think about that picture, how many plans that are being made big and small really do make it to go and reach out and find what our insights saying and what what can they contribute and how can we improve and validate and learn and iterate our plan based on it. And reality is it’s probably not the majority. Of course, it’s hard to quantify because by the very nature of it, there’s a huge unknown component here. But now imagine that with the use of AI and with the use of this technology where we can plug into the assistance, we could turn this around and we could say, instead of waiting for the user to think of asking for the insights, why don’t we tell the AI assistant to think of it and to offer and to proactively do reach out for the insights when contextually in a work stream with the user on a plan, on a topic, it would be relevant? And that is what we try to accomplish here with with this plan… Sorry, with this reviewer agent. And here, I wanna just show you what that can look like. So imagine you’re working or stakeholders working on some topic together with their AI that they leverage. And would, for example, say, I wanna do a launch plan on a certain topic and has already been working on this for a while and has through a longer process come with the AI, maybe to a launch plan artifact, to a PowerPoint, to presentation, to a document, whatever that the AI prepares in cooperation, collaboration with the user based on whatever they have been working on. So far so good. Now what we can do is we can basically instruct the assistant to also think about, proactively think about and bring up, hey, this is a topic that might merit a review from the insights perspective. So after having worked through the actual subject matter plan, the AI can proactively offer here the compilot or the assistant can proactively offer. Here are some elements in what we’ve been working on. We might want to run a review with DeepSights to check it against what we know about consumers and the market. And then maybe I as a user say, yes, of course, go ahead. And then in the interest of time of the demo, this is pre prepared because that might take a while. Then it sends off this plan and the context to DeepSights, which then goes through a structured review process, which uses also, like I showed earlier, all this insights understanding to figure out what do we really need to look at in all the material we have and comes back with a structured review, comes back with a verdict. Does it all look good to go based on what we know from our customers? Is there maybe something to rework, to build on, or there may be gaps also in our knowledge that we need to fill and where we need to validate. And it comes back with a verdict and the recommendation, which can be a high level, excuse me, and which can also then go into more detail and say, here’s exactly different aspects where the review says consider some tweaks, consider some feedback, some gaps, as we compare what you are trying to do with what we know how the market works. And this can then all of course, again, as we always do, give the evidence and the links and the underlying sources. And now the assistant can pick it up again, can take that information, can proactively offer how to then work on the next steps and approve and iterate whatever the person is working on based on what was read you grounded in the insights turned out. And so with that, what we can accomplish is a, proactively bringing up the insights whenever anybody in the organization is working on something that would merit this review. Of course, then ensure that everybody uses insights to the best possible extent to inform their plans. And also at the same time, we can even collect signals in terms of learning what is the organization working on, what insights are required to help them, where are maybe the gaps, where there was a plan and we didn’t have something that might have been helpful to answer an important question that was central to the plan. So this is also a very important source of signals back to us, back to the insights function in terms of what’s the demand and what’s the use of insights in the organization and what’s the topics the organization is working on. So going back to the presentation, what that means is, yes, you can also ask your copilot today to review your plan. But what makes the difference here is a, not only the data that it can use to ground its perspective in, but also, again, be the discipline of how to exercise and run this review. And, again, that is what we discussed earlier. That is what we encode in DeepSights. So you get your discipline, your perspective, and best practices that the AI follows and not an ad hoc review that may or may not be good or helpful. And again, what that means is it gives us… It gives you a way to set a standard, decide what sources to use, how a review should look at certain problems, what are the lenses to apply, what are the verdicts to give to then configure that, to then use the technology to apply it, to to offer it to every plan people are are working on with with the AI infrastructure, And then to use these signals that come back on what is being planned, where the gaps are, what the research and the insights are that really help validate or falsify or challenge plans to use that to also show what is the impact of the insights to the real work that’s happening. And so the breach of the insights grows through the AI without having to have more headcount that would then be essentially sitting with stakeholders directly working through this. So we believe this is a very powerful concept to help scale the use of insights in the organization. And now I wanna give it back to Nick quickly for the fourth use case, the last one I wanna talk about, which goes more into the direction of also dedicating and for the work to the agents. Okay. Thanks, Olaf, again. So let’s… With the reviewer, as Olaf covered, basically, what we have seen is, that you can use this agentic component to review a plan and, essentially validate it against your sources. But this is not the only, evaluation that DeepSights supplies via HMTP. Soon, we will also have the ability to, let’s say, test any ideas you may have with your synthetic personas. And just to make it a bit more clear, I guess more people will be more closer to synthetic personas, but those are synthetic consumers you’re creating, which which are research backed by the evidence that we have in DeepSights. But imagine you do have an idea, and the idea is that this specific case, you would like to test what your consumers… How their consumers will react on a very specific type of campaign line. So you wanna test this against your consumers. So via Copilot, we allow this type of capability that you can delegate this task over to DeepSights. DeepSights will run this specific line by all your configured personas. Again, those personas, are basically created based on your segmentations. They’re created based on your evidence, and then they provide also some sort of a verdict verdict back. They basically tell you, is this a good idea? Does it need to be adjusted? And then you’re getting a little bit of a better understanding who of your, let’s say, different persona types like your idea or not, and then what needs to be refined in the overall plan. For the sake of time, we wanted also to, show you a quick demo, but those authentic capabilities run for quite some time. But I wanted to basically give you the, ability to see how Copilot thinks, real time when you’re delegating the task and then thinks about all the next steps that need to be followed in order to validate your concept against your personas. And after we see the video, then we will pass it back to Olaf for the closing remarks. So we are back in the same conversation with… That Olaf started with the reviewer. The plan has been reviewed. The deck is updated with six changes, and Copilot offers to test the new line we chose with our personas. That’s now connected to DeepSights, so it basically knows the capabilities we offer. On click, you can see Copilot checking what’s available and loading the DeepSights persona scale to use. And these are not generic personas. They are configured on our specific type of segmentations. So what happens now? It’s essentially running a quick focus group on the plan. It shows the line we chose to wellness seekers, value seekers, treat lovers, and each one reacts based on the research it’s built on. And here’s the lee… Read. It lands with the wellness seekers, which is also our launch target, and the show what tells me straight away what still needs a little more work. Its segment gets a short summary and a couple of quotes forced all cited back to, the research is coming from. Treat lovers, as an… As we can see right here, find it a lit… A little bit like a supplement, so they want flavors first. What’s next? Again, it doesn’t stop at reactions. It tell us what are chains and offer to update the deck or rerun the review to make sure that we’re not, essentially running any, ungated plan back to the organization. Yep. Yep. And, yeah, again, that pretty much summarizes overall how personas can also offer some time of, I would say, evaluation against your your test. It runs focus groups, provide the final verdict, and then ball again is in your court. Do you want to re rerun the review? Do you want to refine your campaign line in order to align it with the organization, I would say, I’d say, direction. Yeah. So passing passing it back to Olaf now for his remarks here. Yeah. Thanks, Nick. Thanks for sharing that use case as well. So to wrap it up, what we saw was first of all, how we can embed DeepSights, how we can embed insights to allow users to ask questions right there, how we can use it to also bring together a multitude of other partners and their capabilities into this same interaction. We saw how we can turn the capability into something proactive that actively offers to review content that is being worked on. And we saw also how, for example, personas and other capabilities can be used as active components, be it here in this assistant workflow or be it being triggered by other agents that your company is building to solve certain problems. To wrap it up, as Nick said, I just want to turn it around for a moment and look at it from the business perspective. What would you say now as a business stakeholder? What does that all help me with? I mean, it’s great to have all these integrations, but what’s the so what here really? And I think there is different angles that you can look at this depending on who the stakeholders are that can be around, of course, accelerating responses, reacting faster, having everything integrated at the finger tips with reliable interpretation. So react in hours, not weeks. It can be about winning on the proposition, making sure again, that also in brand and comms that, for example, again, all the best knowledge that we have about consumers and perceptions is integrated right there in the flow. It can be the bigger perspective of packing the right bets when it comes for example, to innovation and R and D also there to validate and ensure in those flows insights are first class citizen. And ultimately, across the board, what we see in blue here, no plan shipping blind in the sense of everything that is done in the organization that requires an insights perspective, get an insights perspective, regardless of whether people may be actively think of it themselves or not. And what does that mean overall? I think overall, what we believe and what we see with our customers is that really this AI enabled, AI first organization is taking shape. There’s a lot of initiatives. There’s a lot of thinking how to structure not only the technology, but also processes and workflows around AI. And we really strongly believe this is now the moment that gives an opportunity to make insights the stewards, make the insights function the stewards of this discipline of understanding the market and consumer at the outside context. So that the insights functions really owns and defines what counts, what are the sources, how to look at it, how to interpret it, what are the rules, what’s the discipline, and that then gets, of course, implemented and connected by the IT teams and amplified through the AI so the insights really become part of the work everywhere. And where that leaves us at the end, of course, also with a call to action, if you’re interested on the customer side, please do reach out. Let us sit together. Let us look at both what… What’s your assistance? What’s your infrastructure? Looking together at what does it take technically to bring it to life and also doing a clear AB side by side comparison to really prove also what we’ve been talking about earlier, that our encoding of the insights discipline basically in DeepSights makes a big difference versus having a generic AI just trying to give an answer based on documents. And for any partners, of course, also here, please do reach out to to join our partner program to make sure we have an out of the box streamlined integration capability that we do offer our joint customers the best possible answers based on the variety of the expertise and data that that we can offer and bring together. So with that, I wanna close it. Of course, we do have some time for questions. If there are any questions, and other than that, I will also be super happy to then get in contact afterwards to speak a little more detail over video demo or whatever. Alright. So that’s alright. That’s final one. So any questions? Do you have something for us? You can see Maria. Great. Yes. Thank you so much. Thank you so much, Nick and Olaf, for a great presentation today. We do have some questions. We do have around five, seven minutes for that we can offer today. So let’s start with the first one. If an insight if insights team is just getting started with AI, where should they begin? That’s a broad ground question, an excellent question. I think, first of all, of course, it needs to start with making sure that, again, the government’s single truth basis is is in shape, but that’s not necessarily by AI question per se. I think if it is about getting started to tap into this ecosystem, become part of the ecosystem, it’s about understanding a… How to how to how to ensure that whatever your AI process exposure to the wider enterprise is something that implements your best practices, your perspective of governance that you trust to, in a sense, speak on behalf of you. So that deserves a very deep review and a very conscious exercise, because ultimately, in a sense, the AI becomes kind of representative of you. Great. Thank you. And then I see that’s an interesting one. What is the biggest mistake you see companies make when they bring AI to insights into the organization? Well, we may have skinned the game in the perspective here. But a common concept, I think, is that there are big AI initiatives going on in companies. Often times they are driven from a specific business process, marketing processes, for example, and then insights aspects get kind of bolted on as an afterthought. And then it happens here and there that maybe also the teams that are responsible, for example, for implementing those AI initiatives think that, yeah, insights are difficult. Can it be? Let’s just throw those research documents in and then the AI will figure it out. That’s maybe because at the surface, it looks simple, but as we know, it is quite a complex and most field. And then… Yeah. Getting that right, making clear that there is much more to giving the right answer than only having access to the documents and getting that anchored and that that perspective in that that is… Can be difficult and can sometimes only be learned the hard way. Got it. Thank you. So not making it the afterthought. And Yeah. And then we have another one that is very… Oh, if you wanted to add something. No. No. I was basically just wanting to echo what Olaf was saying because what we also hear when we discuss with many different customers around AI, like, the most most of the times people think it’s just the LLM, basically. You’re connecting with your different types of subscriptions. But for my experience, the most important thing is, one, understand the outcome. What is what I’m trying to achieve? And the same time, think about the context. Context is becoming, let’s say, the token of the exchange of AIs. Essentially, it is becoming the most important thing to consider while instructing your AI. So the main thing I wanted to emphasize here is that the AI instruction or in other words, how you teach your AI to make certain decisions even in the insights organization is something that it’s not gonna go away go away anytime soon. It re… It reads… It needs some sort of, let’s say, careful attention in a very frequent manner. As again, use case is key to evolve moving forward. Thanks, Nick. And then we have another practical question. So what does IT actually need to do to connect this and how long should it take? I think that’s one for you, Nick. I can I can also take this one? So this is actually a very quick process. Essentially, all of the things that we do, let’s assume that you do have already access to DeepSights because I will just basically group them into two different types of, I would say, cohorts. Cohort number one, you do have access to DeepSights. All we have to do there is we supply you with a key. That’s why on the MCP server, you provide us with your callback URI. Again, this is just a single link that we need. And then based on your DeepSights authentication, basically, the authentication goes through. You have access to the MCP. So this is a little… The, let’s say, the easiest approach. We also have, let’s say, the cohort number b, like the second one, which is you don’t have access to DeepSights. This specific type of, use case is a little bit more, I would say, special because what we have to do is we need to create what we call a technical user, like a user that kind of, let’s say, acts like your internal user, and then we have to define together our, let’s say, permissions, what this user should see, and this is basically how everything is respected on the insights that we bring over to your agent. And the same process has to follow there. But the only… I mean, the main difference is the, let’s say, technicality of it, like how we can create this new user for you. And, again, I don’t wanna downplay it, but at the same time, we may need to have a little bit of a conversation exchange discussion with your agent owners because we can supply with skills how to best harness the capabilities of our system. But at the same time, we also need to understand a little bit the, let’s say, AI landscape or maybe the MCP landscape of your agent to best instruct when it makes sense to use DeepSights the most or get the most out of it. Again, it’s a little bit of a expert exchange on on this particular angle. But it should be quite straightforward from, I would say, the most IT technical experts on the organisation to make the connection happen. Thanks, Nick. Very comprehensive answer, I think. And then we have time for two more. So where do you see this going in the next twelve months or half a year? Where I see well, I mean, I see it going basically the direction we have tried to lay out. I think, again, I believe this is now the perfect time to make sure these structures respect insights as a first class citizen in the ecosystem. I expect still, of course, as we all know, a lot of change and disruption anyway because of technical changes. Also, many organizations are still experimenting and so there will be a lot of uncertainty still. And I think ultimately it will take another, I don’t know how to guess, but it will take more than twelve months for the dust to settle and then to finally come out with what the architecture of the future will look like. Great. Then we meet in twelve months and discuss this topic again. So the last question, this is for me just for you both. If people were to take one thing away from today, what should it be? For me, it’s… But I’m repeating myself to death now, and I’ll keep it very short. It is… This is an opportunity and not a threat. Of course, if we don’t do anything, it may be a threat because then other players and the ecosystems will just say, yeah, whatever. We will solve it even though they may not know exactly what it means. But there’s a great opportunity to make insights part of this and to make the insights function the steward that defines how understanding of the external world is anchored in the AI world. From my perspective, would just wanna say this is an amazing learning opportunity for the majority of the organizations. Like, this is a journey. This is not something you will learn in a single day. And at the same time, again, always think about the outcome. And at the same time, you have all those different types of subscriptions. Think how they can bring the most value on the enterprise tools you use. So let’s find the most valuable space to take into this, let’s say, evolution journey. Great. Thank you so much, Aleph and Nick, and thank you so much for everyone joining today. We will share with you the slides and the recording and the poll answers after the webinar. Please make sure to come back and get in touch with us if you have any further questions. Thank you so much, Aleph Koenig. Thank you very much. Have a great day. Bye bye.
Description
Enterprise AI is being wired into how global companies work. Copilot, ChatGPT, Claude, Gemini and a growing fleet of internal agents are now where employees ask questions, draft plans and make decisions. Yet most of these tools were built to understand the organization, not the market it competes in. Ask them about consumers, categories or competitors and the answers are often unsourced, incomplete or wrong, and they travel fast.
Meanwhile, the research that large organizations spend tens of millions on each year sits one step removed from where the work now happens.
In this masterclass, Olaf Lenzmann, Chief Innovation and Product Officer at Market Logic Software, sets out how leading enterprises are designing the flow of market intelligence through their new AI operating models. He shows how insights and IT leaders can work together to give every assistant and agent access to trusted, governed and cited market knowledge, without losing control of what counts as true.
The session is built for insights leaders who want their research to shape more decisions, and for IT and AI leaders responsible for designing how knowledge travels through their enterprise.
What you will learn
- Why enterprise AI needs a market intelligence layer, and what happens to decision quality without one
- The three patterns through which intelligence can reach more people and agents in an AI-enabled enterprise: ask, review and orchestrate
- How to govern trust at scale: who decides which sources count, how permissions carry through, and how to make sure agents consult your research before the open web
- A practical operating model for insights and IT, including how to start with one critical connection and build from there
Speakers
Olaf Lenzmann, Chief Innovation and Product Officer, Market Logic Software
Olaf leads product strategy and innovation at Market Logic Software, including DeepSights, the company’s AI-powered market intelligence platform. He works with insights and technology leaders at global enterprises on how AI is changing the way organizations find, trust and act on market knowledge
Nick Papakonstantinou, Senior Product Manager, AI and Innovation Portfolio, Market Logic Software
Nick helps shape the AI capabilities behind DeepSights, focusing on making market intelligence more accessible, reliable and actionable. His work spans AI agents, knowledge retrieval and evidence-based decision-making, helping enterprise teams move from questions to trusted insights faster.