The Anti-Copilot

From Institutional Knowledge to Business Intellectual Property

Institutional Knowledge in Industry

If we imagine automation as tying processes together with tight wire, institutional knowledge is the necessary slack in the system that keeps fickle connections from breaking. While more and more human work is abstracted, the importance of those tasks that require human knowledge only increases. Here lies a vulnerability.

Institutional knowledge is the knowledge or wisdom developed that is unique to an organization. That process that no one has ever written down but you can always shoot Sally a Slack message about when you get stuck, that’s institutional knowledge that Sally holds. Or when that valuable, long-time customer that is a bit finicky but Joe always knows the right thing to say, that’s institutional knowledge he is leveraging. Or perhaps that “gut instinct” that Sue has about estimating the cost of a new project, she’s probably leaning on her institutional knowledge in doing that.

Particularly in many foundational industries, such as manufacturing, construction, healthcare, and aerospace/defense, among others, this institutional knowledge is both a valuable commodity and business risk.

It is a valuable commodity in that, on an individual level, having institutional knowledge gives you a leg up on your colleagues. In a similar vein, on a company level, in many industries, having a concentration of institutional knowledge within your workforce can help you beat competition.

At the same time, institutional knowledge is a business risk for a company. It leaves the company vulnerable to turnover in key employees and increases the cost of labor given the importance of particular roles that depend heavily on institutional knowledge. In a different vein, if the aforementioned Sally and Joe have the same job at the same company but they depend on their stored knowledge and “gut”, their output will often differ. This represents a challenges towards automation, which often requires standardization.

For these reasons, and for the sake of ‘constant progress’ in our path towards further augmentation and automation, there will be significant money to be made by using software to digitize and standardize what was previously held in the minds of humans as institutional knowledge.

A Proposal: The Anti-Copilot

But what might this look like? One theory is many roles will be joined by the ‘anti-copilot’. Increasingly, our actions, both in the digital and physical world, output data. Put crudely, they are tracked and quantified. Even if we are not writing things down or doing the thankless work of documentation, through data, computers can do so on our behalf. To this end, an anti-copilot would be a platform that records, digitizes, and documents everything from workflows to communications utilizing computer vision, large language models (LLMs), and multi-modal models on text, video, and audio.

For instance, let’s pretend I am a project manager for a big engineering project, such as building a bridge, tasked with addressing complex regulatory requirements while managing long-standing stakeholder relationships. Increasingly, the legal agreements, regulatory filings, and legislation I might quickly reference or know off the top of my head can be digitized, searchable by a computer. My communications are done through text, email, and chat, all of which can be ingested into an LLM and understood so well they can be emulated. Through this, an anti-copilot would benefit from my last Slack message to Sally, the source of institutional knowledge, about the process around a specific zoning rule. What I see with my eyes on site, increasingly is also documented in photographs or event recorded live by cameras as what is processed by my primary visual cortex in my brain can be done by a convolution neural network running on a computer. An anti-copilot would benefit by working alongside the aforementioned Sue last week, learning her gut instinct for estimating cost at first sight for the benefit of my assignment this week. Through this, her order of operations, live troubleshooting, treatment of relationships, and reference knowledge can be transferred from institutional knowledge to business intellectual property.

Why We Need a Future without Institutional Knowledge

Perhaps this all sounds a bit dystopian. But digitization of institutional knowledge is inevitable and important to the continued running of major industries. As you might have experienced in your own workplace, such stored knowledge is often crucial to core competencies of a business such as troubleshooting a frequent and thorny issue or quickly doing a process that takes peers multiples longer. The Bureau of Labor Statistics estimates that a quarter of the U.S. workforce is over the age of 55, and of these, a third are 65 or more [1]. As these individuals age out of the workforce, immense skill gaps will be left in their wake. According to Korn Ferry, by 2030, more than 85 million jobs globally could go unfilled because there aren’t enough skilled people to take them. That talent shortage could result in about $8.5 trillion in unrealized annual revenues [2]. Given these societal trends and economic impact, we must augment human labor not because we can, but because we must for the continued expansion of the global economy.

So, next time you are impressed by your colleagues' instinct in the work place, appreciate their knowledge and perhaps consider what your job might look like without it.

?? Read more

[1]: BLS via Fortune

[2]: Korn Ferry