Lucy the AI finds diamonds
In Part 1 of this article, we sat down with Steve Frederickson, Director of Product Management at Lucy to talk about how generative AI within a company’s network of information points can cut down wasted search time exponentially. Lucy produces an eponymous AI tool that searches a company’s structured and unstructured data to put answers quickly at the fingertips of staff in large enterprises, saving them, their colleagues, and the company as a whole the time and hassle of searching though potentially siloed data stores, day after day instead of doing their jobs.
In a sense, Lucy finds diamonds in a coalmine of potentially impenetrable data, and it does it in a fraction of the time it would take either staff or a standard Google-style search.
The addition of generative AI.
In January, 2023, Lucy updated its offering with a generative AI element, called Lucy Synopsis. We asked Steve, apart from the hot-button chic of generative AI in the year of ChatGPT and its sudden flurry of competitors, what the new element brings to enterprises with a lot of data locked somewhere in their bones and silos.
Here’s a question. What is an answer?
That’s a very philosophical way to start.
An answer is a unit of knowledge. If you’re in an enterprise, you have lots of answers to questions as yet unasked, locked up in the data of your day-to-day existence. So when you ask a question, you can be pretty sure the answer’s there somewhere. And we’re always looking to get you closer to the core elements of content that you’re looking for, that will give you the answers you need.
Generative AI is an amazing supplement to that, because here we have those specific answers. Here’s the ten things that are suitable in your content that represent your company’s knowledge about this topic. Plus, to help you understand that, here’s a summary of that content, here’s something that you can take elsewhere.
And so, by taking the best of both worlds and building an experience that is beyond the blue link, but also respectful of the limitations of AI and its current state, we believe that we have a solution that is the future of not only enterprise, search and knowledge management, but something that is truly empowering for people. Helping them find what they’re looking for, go back to their day, start those conversations, and unlock that data.
Mining for diamonds in the data.
That’s the thing that won’t necessarily occur to people about generative AI within an enterprise context. The idea of mining unstructured data, so as you say, you can ask a question and get to the minute of a video that was maybe shared between three people six years ago, and yet within that minute is the answer you need, and that would maybe only be discoverable with certainty in that minute of video. It may have become part of the mental knowledge of maybe 12 people in 10,000 since then, but how long would it take to find that out? Instead, boom! Thank you, generative AI, I’m off for a cup of coffee.
Totally, totally. There’s just so much content within any given enterprise. In the consumer space, you have Google, and Google’s magic trick was to get everybody in the world to tag their content and make it available for search.
But if the answer you’re looking for is in a PowerPoint somewhere, or a video, or a SharePoint file – you’re never going to get an enterprise that tags all of those.
It would more or less take up the day’s work-time just to tag everything.
Exactly. It would take millions of man hours to go through that experience. So here we have this incredible index that we’ve built using the power of AI. And then on top of that, we have the best of both worlds, with answers and generative AI at the same time.
Better problems, better solutions.
We’re always looking for new and better ways to represent problems, but how do you ensure that the right solutions get picked and that they get tailored in a way that is thoughtful and meaningful? What we do is establish a set of principles. We believe that software can be empowering, but we believe finding answers can also be delightful, and we want to create those moments of joy. We believe that AI is here to empower people, it’s not here to replace them.
And we think that this experience is helping people reach higher using the data that they find. We want to put your content center stage, prioritizing your content and the people that authored it, so that we can help people get back to their day, save them time and teach them things that they didn’t know existed. And that’s something that we get very excited about every day, so we built that into our process in our system.
Give us an example of a moment of joy experience.
My favorite example is when I ask Lucy a question that I don’t think we have content for, like it’s not connected to our sources, maybe it’s outside of our ecosystem of knowledge or something. And I find a document that I didn’t know existed. And then I’ve got this great content, this answer from the answer engine, but then Synopsis, the generative AI element, gives me a meaningful summary of what I’m looking at. It takes that content and makes it portable.
That’s an incredible balance, because then I’ve got the short version, the summary, and then there’s some content that was completely dormant. Nobody’s touched it in however long and it’s in our incident storage, and Synopsis helps me understand it, but it doesn’t replace it. I can go in and unlock that content, and put it in this new project that I’m working on. That’s very exciting for us.
Lucy… on Lucy.
We were in the room with some C-suite people recently, and the question came up of how they saw the future of Lucy in their company. I said I could ask Lucy that question, so we did, and there were all these resources – pieces of dialogue in Teams meetings and the like – that showed the journey to that room, and what they had imagined the future of Lucy in their company to be. It was amazing to have Lucy sort of be a voice in that room.
And arguably the most objective voice, because it’s based on the factual data that has existed on the journey to AI.
That’s exactly right. It was written by a human and an expert, somebody who did something thoughtful in the past and went through this exercise and distilled all that down. And then we have the synopsis response as well, which helps summarize that. Lucy was able to find that data that we needed, in a moment, to answer a question it would have taken too long to find any other way.
Lucy, the AI, found diamonds, in fact?
I can’t believe you just said that…
Generative AI will change the world in almost every way we’ve known it. Lucy – and especially Lucy Synopsis – is a leading example of a way in which it could be used to solve business problems, especially in large organizations with lots of unstructured data, while helping staff get the answers they need to all their important job-related questions.
What the future holds, nobody knows (although clearly, Lucy could probably give you a good idea). But the long-term adaptation of generative AI to serve genuine business needs has to be a niche that’s worth exploring, alongside the more generalized search engine functions being developed by the rival tech giants in early 2023.
20 March 2023