lucy ai screenshot

Lucy


Founded 2015 | HQ Minnesota, US | 40 employees | $4M annual revenues

Lucy is promising to be the KM answer engine for the terabytes of unstructured data found in the typical medium to large enterprise … (and) based on what we’ve seen so far, we think Lucy can succeed as a niche solution for specific use cases.


The Company

Need insight from all your unstructured data? Ask Lucy for the answers.

This is the fundamental value proposition of Lucy, an AI software company with deep experience in natural language processing (NLP) and natural language understanding (NLU). Founded six years ago as Equals 3, an IBM Watson channel partner, the company initially focused on mining advertising and marketing insight solutions for global brands.

In 2020, the company changed its name to Lucy, also the name of its flagship product that was born out of a special solution it delivered for PepsiCo to provide research and insight for the marketing department. Why Lucy? In homage to IBM’s pioneering AI work, Lucy was named after Lucinda Watson, the granddaughter of IBM founder Thomas Watson.

The company has competed historically in the mar-tech market but recently established a basecamp at the foot of Knowledge Management Mountain. It hopes to ride the recent KM renaissance by positioning Lucy as a general-purpose, AI-powered assistant for enterprise knowledge management. The company has raised a total of $16 million from investors, with a Series A round completed in December 2021.

The Technology

Before we look at what this does, first it’s incumbent to note what it doesn’t do. The Lucy team was quick to explain that Lucy doesn’t actually provide “insights” from unstructured data (documents, files, emails, etc.). It is better understood as a cognitive companion, using its AI to sift through all the available data sources, then finding and bringing actionable data to the right person at the right time to inform their business insights. Lucy is not meant to replace existing KM or ECM systems that do some of this; it functions as an overlay on top of any repository or file system where the documents are already stored.

Lucy uses machine learning and NLP to crawl through file sources, tagging and indexing the files while leaving them in place. While Lucy is not the first company to do this, index-in-place is a very useful feature that eliminates the hassle and security risks of uploading and curating all of one’s unstructured data before any analysis can be done. Once Lucy connects to a data or file source, it becomes persistent so that any new content added to the file source will be indexed and ready for queries. Lucy has integrations with popular cloud file storage products such as SharePoint, Google Drive, Dropbox, and Box.

In a typical KM system, subject matter experts were required to spend many hours curating the data and creating tags and labels so the information could be accurately retrieved. The company claims that Lucy is capable of reading, categorizing, and labeling large amounts of unstructured data with minimal human supervision.

Once the unstructured data is indexed and tagged, Lucy leverages its natural language understanding (NLU) power to become an “answer engine.” (Note: the company does not want Lucy to be seen as just another search engine, so they have gone so far as to claim a service mark for the phrase “answer engine.”) You can type in natural language questions and Lucy returns specific answers from the indexed data sources, then shows you the exact spot in the document(s) where the answer resides (see Figure 1).

The product uses machine learning to continually learn and improve as it processes new files. When search results are returned, Lucy shows the machine’s confidence level for the result. By liking or unliking a result, the user trains Lucy to do better the next time.

On the innovation front, we liked that Lucy can search through databases with the same query. You can also connect Lucy to externally licensed marketing research databases such as Nielsen, eMarketer, Mintel, Kantar (Retail IQ or Monitor), IDC, Euromonitor, Stylus, and others. The company has even patented this method for simultaneous query of unstructured and structured data.

lucy ai screenshot
How Lucy Answers a Question

In this example, Lucy answers the question “how are trends in online shopping changing?” by returning a report from the company’s Dropbox as well as reports from external licensed sources – in this case Nielsen Insights and eMarketer.


Our Opinion

We have tracked knowledge management software for almost 30 years and watched as myriad vendors and end-user organizations have tried and failed to implement a true enterprise KM solution. Some progress has been made at the departmental level where subject matter experts live. But as with enterprise search, enterprise knowledge management has been frustratingly elusive to achieve. The recent explosion in cheap AI computing power coupled with easy access to data stored in the cloud may have created the perfect petri dish for KM to flourish.

Lucy is promising to be the KM answer engine for the terabytes of unstructured data found in the typical medium to large enterprise, and that is a big ask for a product that until recently was focused on marketing data. However, based on what we’ve seen so far, we think Lucy can succeed as a niche solution for specific use cases. With caveats: if the underlying AI is as powerful and easy to use as advertised, and if Microsoft’s KM platform does not quite live up to its hype.

Advice to Buyers

Lucy is fairly easy to evaluate because it leaves your files in place, so it may be worth running a proof of concept on a specific corpus of data to test the company’s claims. While the company’s marketing mentions “instant access” to terabytes of data, that assumes all your data has already been crawled and indexed, a preparation process that can take weeks to months depending on several factors.

What can be more challenging, though, is creating the business case to justify the purchase of Lucy or any other KM or enterprise search product to the C-suite.


SOAR Analysis

Strengths

  • Natural language understanding expertise
  • Experts at marketing and advertising data mining

Aspirations

  • Become known as the KM answer engine
  • Grow beyond its legacy mar-tech base

Opportunities

  • Build out industry-specific solutions where the giants don’t play
  • Build strategic partnerships with Salesforce, ServiceNow, and Microsoft

Results

  • Blue-chip global customer base with growing ARR
  • $16 million raised to date

Attribution-NonCommercial-NoDerivatives 4.0 International
CC BY-NC-ND 4.0 license