Taming information chaos to drive successful AI initiatives

For AI initiatives to be successful, a heap of other KM issues—such as those relating to data quality, data silos, lack of resources, and more—must first be tackled. Culminating in a certain “digital mayhem,” AI requires data to be accessible, reliable, and interoperable to generate any sort of tangible business value.

To help navigate these obstacles, Ville Somppi, vice president, industry solutions at M-Files, and Anthony Vigliotti, chief product officer at Adlib Software, joined KMWorld’s webinar, Before You Dive In: The Key Information and KM Issues to Tackle for AI Success, offering a myriad of strategies for enabling a data infrastructure that best supports AI.

“We do live in a world of information chaos,” began Somppi. “More information and more data is being created all the time, and the tools that knowledge workers have to use every day get more complicated.”

With more applications for specific tasks than ever—propelled by enterprise growth and increased software investment—data flows have become entirely disconnected. While generative AI (GenAI) holds promise for reconnecting those data flows, challenges associated with connectivity, access, curation, and balancing costs and benefits must be addressed first.

Somppi explained that to succeed with AI and to free up knowledge workers’ time, automation is required. Important things to automate are:

• Classification and organization of data
• Workflows
• Security and compliance

By automatically organizing information for knowledge workers, there can be about a 50% improvement in information search, according to M-Files research. This ultimately invites ease around finding and using knowledge within the business context.

Similarly, if workflows are automated successfully across the creation, collaboration, and approval of documents and data, enterprises can see 70% greater workflow efficiency. If access rules are automated, accompanied by secure sharing and collaboration, enterprises can enable 65% faster document filing without compromising security.

Vigliotti emphasized that AI initiatives are tied to document management issues, where a document management system can resolve information chaos.

“If you think of an individual that you have in your environment, their mental models are all different…some are really good at organizing their content, some are really bad at naming and organizing their content,” explained Vigliotti. “You may be feeding your potential downstream document management system a very unorganized structure.”

The key, then, is to ensure that information is clean and organized before it ever reaches the document management system. However, this is not necessarily a technological issue; it actually relies on people.

As knowledge is onboarded to an organization, the cleanliness of information—which spans various sources and locations—is entirely dependent on its human interaction. Namely, it relies on people to:

• Properly name and folder each documents while following established business rules
• Understand and comply with external regulations regarding data security and privacy
• Ensure each document is accessible by the organization

There is room for opportunity in this chaotic, human-reliant information mess—specifically as it relates to automation and repeatability.

“What we’re doing at Adlib is we’re taking those sometimes complex, highly repeatable manual tasks in this document origination and automating them by way of ruleset,” said Vigliotti.

As a document is received by an environment, Adlib assembles the document with consistent conventions and accurate access policies and delivers it to the desired location. Adlib automates mission-critical processes with human-like intelligence and machine-like precision for highly regulated industries. By transforming data and documents across an enterprise into delivery-ready materials, Adlib tames the digital mayhem plaguing KM teams amid AI surge.

For an in-depth discussion of KM data strategies to successfully drive AI initiatives, you can view an archived version of the webinar here.