Knowledge Systems Finally Learn to Think for Themselves
It took over thirty years to digitize enterprise knowledge systems. Now what?
In 1983, I packed all my belongings into the backseat of my 1972 Chevy and drove south, arriving 24 hours later at my first post-college job as a network operations center shift supervisor for Ma Bell.
The union went on strike 72 hours later. I was shown a computer terminal and told, “You are managing the backbone network for five states — don’t mess it up.”
I had no clue what I was doing. Bell Labs, however, did. There were more than 40 3-inch to 4-inch manuals stacked on the wall behind my desk. It was my own personal knowledge bank.
A day later, an alert on my screen told me a TD-3 Radio failed. Sure enough, one of the manuals behind me was labeled TD-3 (here’s what the first chapter looked like).
Hours went by as I read through manuals while 10,000-plus people in Mississippi waited for me so they could resume making long distance calls. Eventually, I dispatched an equally clueless strike duty volunteer to the radio station and together we read enough to restore service without electrocuting him in the process.
After the breakup, I took a job at AT&T as a sales technical support manager. At the time, a sales rep was expected to sell network services, PBX/call centers and data networking services. Half of my team were network engineers and half were PBX/call center reps; nobody knew data.
It didn’t take too long for us to lose a $1M account to MCI. We had an upset client claiming that their customers were not getting through on their 800 lines. The sales rep contacted her PBX sales engineer. No trouble found.
None of us, myself included, had ever been to the 800# training class. If we had, we would have run a standard report and quickly solved the problem. The sales executive was expected to know their customer and know who to call. AT&T eventually gave up trying to sell network, phone and data equipment and spun out Lucent (then Avaya). (Note to self: Institutional knowledge is not only about sending, it is about receiving, absorbing and changing behavior.)
Let’s jump to 1998, when everyone had laptops equipped with 3.5-inch floppy disks or CD-ROM drives. Looking back, this was the beginning era of the “dark pool” of bits. More on that in a moment.
I was a GM of a $2B+ division operating in 95 countries and multiple languages. We had e-learning centers, content management systems, web sites and everything else we thought we needed.
Once a year we geared up for a major new product release. We had the “bible” of what new features/products would be launched. Knowledge workers would take from the bible and in turn create a laundry list of stuff, including sales collateral, sales & sales engineering training, installation training, customer training, and trouble-shooting guides. A month before ship date, features started to slip. All of that stuff we produced was instantly obsolete or conflicting. Ever worse, the dark pool of bits took over. Invariably, I would hear a well-paid sales person deliver last year’s pitch from their hard drive.
Fast forward to 2015 and ask a CIO how much stuff is accumulating in their organization. Email threads, SharePoint drives, Dropbox/Box accounts, laptops, mobile phones with gigabyte drives, and tablets with even more. Each repository is stuffed with presentations, sales collateral, manuals, video/audio files and configuration notes. The dark pool has grown to petabytes.
What has changed? Curators still try to locate, find and publish the latest version of knowledge. We discovered “portals”, “SharePoint folders”, and “key word search”, but these are just digital book shelves of the same type that held all of my training manuals from 30 years ago.
Corporations spend $150B a year on knowledge management and training systems. We’ve digitized the manuals and moved the classroom online, but we haven’t made any real advancements.
In the next 10 years, we are going to see knowledge create a mind of its own:
- Machines will accurately curate 80% of the dark pool, enabling discovery of what institutional knowledge is where. Unstructured search algorithms will understand what is really in our bags of bits using clues like who published this document, when it was published, what kind of info is in the document (is it an installation/repair guide for TD-3 radios or is it a warranty form?), what kinds of products are covered in the document. The dark pool is going to get attacked by smart machines.
- Knowledge takes lessons from consumer advertising. Wherever I go online, the web is tracking my every move, learning who I am, what I like and react to, and it uniquely targets me. Why not do the same inside the enterprise? You already know who I am and what I am supposed to be doing to drive value for a company. Knowledge will become aware of itself. It will figure out who used it, how it was used by different types of workers, whether it was liked or not, and start figuring out what else to recommend.
- Knowledge gets APIs to KPIs.
If knowledge becomes aware of itself, how can it know it is making a difference? The enterprise is full of other applications that measure key performance indicators. If you link knowledge to KPIs, now you can track behavior changes, such as measuring if sales reps using last year’s deck outsell their counterparts using a new deck.
Next generation knowledge automation startup companies are seeing the chance to grab a piece of this $150B pie. We’ve bet on Kaybus because they are tackling all three of the major factors above. Other names with similar aims but different approaches include Witty Parrot, Folloze, Jive, and Bloomfire. Companies that create, share, and absorb knowledge that matter to their key performance indicators are going to out-execute those stuck in the past 30 years of history.