#CallForStartups Knowledge Management

Remote work increasingly looks like the new normal as not only the place we work continues to change, but also the technologies, company cultural norms, and modes of interaction. Let’s just call “remote work,” well, work.

One of the most obvious changes is how we communicate: It’s all digital. Whether we’re meeting virtually via Zoom or leaning more heavily on Slack for quick messages, it’s all happening via bits and bytes. While all-day screen time has its drawbacks, it’s also data-rich and opens a space for new technologies.

Before I point to this new space, here’s some context on the world of knowledge management. The rise of products like Notion emphasizes this trend — People need a place to organize knowledge. The “knowledge” could include sales tactics, programming best practices, marketing insights, and the list goes on.

The systematic management of knowledge is not a new phenomenon. Perhaps the first version of a knowledge management (KM) system was the corporate library, which became more prominent around the year 1900. These libraries solved a basic problem: record useful information and organize the information such that it can be referenced later.

This class of software tools earned the name knowledge management systems. Today, KM systems loosely fall into a few categories: intranets/wikis, structured content management, customer support documentation, and knowledge bases. Here’s a condensed market snapshot of tools for KM (VCs ❤️ market maps):

The mode for knowledge capture hasn’t evolved much since corporate libraries emerged at the turn of the century: we document knowledge manually so others can look for it later. While software moved this process into a digital paradigm, the approach largely mirrors how we used corporate libraries (although, companies like Guru and Vowel have made notable advances).

The analog process of collecting knowledge (write) limits the potential value of KM within organizations due to a handful of constraints:

  1. People have to actually write stuff down.
  2. What we should document is not necessarily obvious. Will an insight shared in a meeting be useful to other teams later? Maybe.
  3. There is a reverse causality dilemma: If I don’t perceive the system as useful, I won’t contribute to it. If I don’t contribute to the system, it will never be useful. This challenge is associated with high starting costs — organizations must commit large amounts of time and money to get started.

What about knowledge retrieval (read)? Unfortunately, accessing this knowledge isn’t that much better:

  1. Knowledge isn’t always available where I need it. For example, if I’m a salesperson selling my product to a manufacturer, it would be useful to understand the talking points that would improve the likelihood of a sale. When crafting an email to the prospective manufacturer, will I pause to search for talking points? Probably not.
  2. Words are messy. Keyword search is cluttered, and topical search works if the ideas have been appropriately tagged. In my sales example, if another salesperson has already determined the talking points, those points would need to be tagged in the KM system so that they’re easily discoverable. The manual task of indexing and organizing information translates to poor documentation. While some KM systems index automatically, there is still considerable noise in the search results.
  3. Even in situations where the information has been appropriately tagged, how do I determine which information I should prioritize? For example, if a developer wants to understand payment infrastructure APIs according to what has been implemented previously, how would she distinguish between varying approaches? What were the tradeoffs considered in the previous implementations, and whose approach is most relevant?

In light of these limitations, here’s a set of design principles that could characterize an ideal KM system:

  • Seamless collection — knowledge capture happens automatically wherever it’s generated
  • Intelligent indexing — knowledge is automatically tagged such that it’s easily retrievable
  • Ambient discovery — information surfaces when it’s needed wherever I’m working
  • Obvious hierarchy — it’s clear which information is the most relevant

Back to “working” and how it opens the door to new technologies. Because most of our communication now occurs digitally, the information that transacts over these channels can be captured, indexed, resurfaced, and prioritized automatically. Thanks to artificial intelligence (AI), the read/write process is seamless. If my colleague shares something useful during a Zoom meeting about selling to manufacturers, I can now reference those comments later to assist with my sale.

Collecting this knowledge doesn’t require a cold-start either. Many companies already have rich data sources of knowledge:

Imagine if all of the information generated in these tools were combined with the dialogue that takes place via virtual meetings such that it could be referenced later. It’s possible to take this one step further by ranking knowledge according to who said it. This new system could observe who people go to for specific topics.

This feature would be analogous to how Google built its search algorithm by prioritizing search results that had other pages linking to those results. What if this system could surface information wherever I’m working? As I’m writing my sales email to a manufacturer, the talking points surface automatically in my email tool.

There are privacy implications of this approach that should be forefront to the design of such a system, but I believe brilliant entrepreneurs can make this possible. In the workplace, knowledge can be ubiquitous.