Lenovo AI Library Validation: Results on Enterprise AI Knowledge Management

Enterprise knowledge management isn’t a new idea. Companies have tried wikis, enterprise search, and intranets for years. But most of those systems were built to store information, not support how people actually work. The result is a familiar problem: employees are flooded with information, yet still spend too much time searching for answers instead of making decisions.

Ryan Shrout and Mitch Lewis sit down with Sarah Lundgren, Director of Technical Enablement for Hybrid Cloud and AI at Lenovo, to walk through Signal65's independent validation of the Lenovo Knowledge Superagent, built on the Lenovo XIQ agent platform. The conversation covers what makes agentic AI a structural improvement over prior knowledge tools, how Lenovo deployed the platform internally before taking it to customers, and what Signal65's testing found across deployment speed, retrieval time, employee adoption, and productivity impact.

Key Takeaways Include:

• Most enterprise knowledge projects fail long before employees ever use them. Signal65 found that the Lenovo Knowledge Superagent can be deployed and operational in approximately two weeks. Custom knowledge management builds run six months to a year or more on the low end, with no guarantee of the governance, feedback loops, or lifecycle management that come built into XIQ.
• Enterprise knowledge overload is costing more time than most companies realize. Signal65 validated a 30% reduction in search time, while employees relied less on subject matter experts just to find answers, with 81% of employees reporting measurable time savings. 94% of users said the agent helped them overcome blank-page syndrome, and subject-matter expert outreach dropped by 35%, freeing up high-value time across the organization.
• The ROI of enterprise knowledge management gets very large, very fast. At scale, the productivity impact translates to approximately 120 hours saved per employee annually, with up to $17 million in potential productivity value in a 3,000-person organization.
• Most AI pilots fail because adoption is treated as a rollout problem instead of a product problem. Lenovo deployed the Knowledge Superagent internally first, building onboarding, feedback loops, and change management into the platform from day one. Clear use cases, content owners, continuous feedback routing to IT, and a diverse early user group allowed the platform to scale across business units organically rather than stalling in pilot.
  
  

Sarah Lundgren's read on where this goes: people will no longer start from scratch. Agents handle retrieval and synthesis so humans can focus on judgment and decisions. The Knowledge Superagent is the first in Lenovo's AI library of domain-specific agents, with retail, operations, and services agents designed to work as a connected system on the same shared platform.

‍

View the full Signal65 validation report at https://bit.ly/3S22IZo 

Watch the full conversation at sixfivemedia.com and subscribe to our YouTube channel so you never miss an episode.

Disclaimer: Six Five Media is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.

SARAH LUNDGREN: We treated adoption and change as part of the solution. Alongside the knowledge super agent, we applied the services that we help our customers with as well, but we used it internally. Clear use cases, lightweight onboarding, and continuous feedback, and made the feedback very easy to provide. That helps us to scale beyond the pipe.

RYAN SHROUT: Hey everybody, welcome to another Signal 65 Video Insights. I'm your host, President at Signal 65, Ryan Shrout. Thanks for joining us today. I'm joined once again by Mitch Lewis, who is one of our lead performance analysts at Signal 65. Mitch, good to see you.

MITCH LEWIS: Yeah, always happy to be here.

RYAN SHROUT: And we are joined by our guest today, Sarah Lundgren, who's the Director of Technical Enablement for Hybrid Cloud and AI at Lenovo. Sarah, thanks for joining us.

SARAH LUNDGREN: Thank you for having me.

RYAN SHROUT: Now our topic for today is we're going to talk about the Lenovo Knowledge Superagent and a recent report that Signal 65 has published looking at and kind of validating the capabilities and performance and value of this Knowledge Superagent. This is built on Lenovo XIQ agent platform. Right. And so you'll see Signal 65 and working really closely with Lenovo on on some of these different proof of concept and kind of product offerings and really showing the value and bringing some of that that knowledge and proof of value into into the marketplace. One of the areas I wanted to start with is kind of just setting the stage about what this is and why this is a product. that people really need, right? Knowledge management as a term is kind of one of those top AI use cases across the enterprise. You know, even Futurum Research, one of our sister companies, right, puts it as second as more than 50% of organizations kind of want to target that use case. Why is enterprise knowledge such a natural fit for agentic AI in your view and in Lenovo's view?

SARAH LUNDGREN: Knowledge is a natural fit because this is where friction quietly compounds every day. People are overwhelmed with information, not short on it. Agentic AI works with the intent and context, helping people move forward, instead of searching. You validated this with a 30% reduction in knowledge retrieval time, basically freeing up people to focus on judgement and decisions.

RYAN SHROUT: Yeah, it's, it's, it's incredibly powerful. I am curious, though, like, this is something that has attempted to be solved many times in the enterprise. I remember, in my time working in corporate America, right, there were wikis and enterprise search and intranets and things like that. Why did those other approaches fall short? Or why are they kind of fundamentally improved on by something like the Knowledge Superagent?

SARAH LUNDGREN: Those tools stored information, but didn't support how people actually work. With agentic AI and drag, the experience becomes conversational. Employees ask questions naturally and get grounded answers with citations.

RYAN SHROUT: Yeah, I think that citations is an interesting part. I want to start at the base though, and ask you about the Lenovo Knowledge Superagent. Can you walk me through what exactly it is, how it fits into this broader AI library of the XIQ agent platform that Lenovo is building?

SARAH LUNDGREN: Yes, so the Knowledge Superagent is an enterprise AI teammate for KnowledgeWorks. It connects securely to systems like SharePoint, Confluence, and S3. It's built on the Lenovo XIQ agent platform, which also powers the broader Lenovo AI library of domain-specific super agents.

RYAN SHROUT: I know the platform supports the open AI kind of API compatible models and it's flexible, right? It can offer cloud hybrid on-prem solutions, which I think is really important. In your view and kind of talking with and working with these enterprises, how important has that flexibility of the offering really been?

SARAH LUNDGREN: That flexibility is critical. Enterprises operate across mixed environment with different compliance needs. The Lenovo Hybrid AI is all about choice by design. So meeting our customers where they are, allowing them to deploy AI where and how it makes sense for them and for their business.

RYAN SHROUT: Got it. Now, one of the interesting things is I'm going to, Mitch, I'm going to pull you in here in a second and start to talk about some of the results that we found and some of the interesting data that we found. But I'm also curious, Sarah, if you can talk me through the rollout of this inside Lenovo, right? I love this idea of Lenovo powers Lenovo. It's something I've talked with leadership about quite a bit there, where the company deploys these platforms internally right, to solve real business problems, you get both the benefit of being kind of first to adopt this technology, you also learn the ins and outs from it. So you know how to help customers utilize it going forward. What was the problem that you were trying to solve? Who were the first users internally? What, anything interesting that stands about, stands out about that kind of Lenovo, powering Lenovo initiative?

SARAH LUNDGREN: I think that it's a very typical enterprise problem. The fragmented knowledge, the low confidence in information. Early users were teams dealing with complex, cross-functional content. Once they experienced faster access to trusted answers, adoption spread organically, We also included a broad, diverse set of key users early on to iterate fast and to capture real feedback. In addition, we deployed our AI adoption and change services to support the larger change.

RYAN SHROUT: I do think that's interesting, the idea of kind of a diverse group of users. I think a lot of times these things start out being utilized by the people that have developed them almost exclusively for some period of time. And you kind of, it's easy to misinterpret some of the positivity and things like that until you get it in the real world user's hands. Now, again, a lot of times these types of things still stall in a pilot form of some case. We've talked about it a ton. here on these video insights, right? Things that go from pilot into production. What did your team do differently this time to move from pilot to a broadly adapted platform or adopted platform rather, across multiple business units and very quickly, I might add too.

SARAH LUNDGREN: We treated adoption and change as part of the solution. Alongside the knowledge super agent, we applied the services that we help our customers with as well, but we used it internally. Clear use cases, lightweight onboarding, and continuous feedback, and made the feedback very easy to provide. That helped us to scale beyond the pilot. because it became a natural part of how we work and it fit very well in because it was across the enterprise.

RYAN SHROUT: Interesting. One of the other features that I know Mitch pointed out and that I saw that was really interesting is this idea of a feedback loop, right? Where users can rate the responses that the agent brings back. And that feedback actually gets routed to IT if there's some technical issues, to product management if there are more kind of like functional level issues. How did that shape the quality of the agent over time? Did you guys see big changes and improvements?

SARAH LUNDGREN: It makes quality continuous and I think that that is key, that we always have a pulse on updating content, making sure that our users feel that they are part of it. So the feedback flows as you described, right? It goes to the relevant parties, if it's IT or if it's the content owners, and it improves our responses continuously. And I must say, because of the ease of the function that we've put in, the users provide feedback all the time. It can be good, it might not be good. And that's what we want, right? We want the teams to provide us information that we can action.

RYAN SHROUT: It's one of the areas I think It's incredibly interesting for you to watch even as we deploy different AI systems internally, as well as that there is this kind of constant and frequent change. And the idea of kind of deciding on some software and sticking with it for three years and its outcome is kind of just not something that is going to happen anymore. All right, Mitch, I want to talk about the measured results, what the numbers showed, what we spent our time with on this knowledge super agent. Can you walk me through some of the key highlights of what you're able to observe and witness through our time with Lenovo on this platform?

MITCH LEWIS: Yeah, so we observed the platform, what it does, the capabilities, and also looked at actual deployment of it. So I think there's a couple of key things, the first one being, this is something you can get up and running very quickly. So, you know, we found that, you know, this solution is something that can realistically be deployed in around two weeks, right? So when you compare that to what are the alternatives of go build something yourself, it's very custom, it's very complex. You're looking at, you know, six months kind of on the low end up to, you know, maybe a year plus. So that's kind of the first piece of like, how quickly can you actually get up and running? And then when we were looking at, what does it actually do? We found around a 30% reduction in time spent searching through various knowledge sources. And then 81% of employees were reporting these time savings. Basically what we found was you could get up and running quickly, it was actually impactful, and people were actually using it. And I think those are, you know, really some of the big hurdles that organizations are facing when they say, oh, AI is cool, how can we implement it, but it takes forever, or, you know, it's not actually a great solution once you have it, people aren't using it.

RYAN SHROUT: Yeah, one of the things that you that you wrote up in the report that I really liked is this, this idea of blank page syndrome, which I'm guilty of myself, right? You're staring at a blank page, you're trying to figure out, how am I going to build this report? How am I going to build this customer deck? And it was like, was it 94% of users said that the that these this knowledge agent helped them overcome that blank page syndrome. And it also resulted in 30, 35% fewer outreaches to subject matter experts, which frees up everybody's time. It just makes the whole system more efficient across the entire organization. And I think there are two numbers, Sarah, that stand out to me that I'm going to get your take on. One of them was, look at the paper here, roughly 120 hours saved per employee annually and up to $17 million in potential productivity value in a 3,000 person organization. What does that mean to you? What's that mean as you go try to position these types of solutions to enterprise?

SARAH LUNDGREN: I think that the biggest impact is time returned to employees. The 120 hours saved per employee annually, which It shows up in faster proposal, better execution, and improved quality of work and employee experience. It helps the organization to scale things faster. It amplifies the job that needs to be done. And I think that those things are really the ROI that any company would really want to have.

RYAN SHROUT: Yeah, I mean, it's the value proposition of this entire solution, right, from, to a, to a business and to an enterprise perspective. I think those, those are very powerful takeaways from it. So I want to, at the end of these interviews, I always like to ask a couple of forward-looking questions. I'm always trying to see what's, this is great. What comes next? You know, knowledge is the, is the first super agent in the series, right, that we've kind of talked about and looked at here at Signal 65, but there are others, you know, retail and other ones coming down the pipe, even from our own projects and analysis. How do you see these types of domain-specific agents working together and kind of what's next on the Lenovo roadmap in that way?

SARAH LUNDGREN: We see these agents as a connected system. Knowledge supports retail, retail informs operations, and operations support services. The AI library, allows the kind of step-by-step adoption on a shared platform.

RYAN SHROUT: What about two to three years out? How do you expect the role of enterprise knowledge workers to change as agents and systems like this kind of become the norm and standard? It's something that's on everybody's topic of discussions these days.

SARAH LUNDGREN: People will no longer start from scratch. Agents will of handle the retrieval and synthesis, allowing humans to focus on judgment and decision making.

RYAN SHROUT: I think it's really compelling. It's something that even inside our own company at Futurum and Signal 65, all these different areas, we've seen a lot of interesting and unique cases about using these types of agents to prepare and prep and kind of validate a lot of the work we're doing. So I'm looking forward to seeing what changes. I'm looking forward to seeing what else comes from Lenovo through the XIQ agent platform. Sarah, thank you for joining us today. I really appreciate it.

SARAH LUNDGREN: Thank you for having me.

RYAN SHROUT: And for everybody else, make sure you go to Signal65.com. Check out our full report on this platform and others from Lenovo coming up very soon. For now, I'm Ryan Shrout for Signal 65. We'll see you in the next one.

‍