Let’s start by defining organizational amnesia, a phenomenon that has become all too familiar for many organizations today. Organizations are losing institutional knowledge due to large-scale layoffs. Since AI went mainstream, the problem has only compounded in volume and velocity as companies opt for AI systems capable of running middle-office and operational functions with fewer employees. However, layoffs without a proper transition plan to capture years of institutional knowledge significantly risk an organization’s ability to succeed with AI.
AI without context can be confidently wrong and massively disrupt business operations previously led by humans. And without institutional knowledge, organizational amnesia sets in, despite the availability of Large Language Models (LLMs), strong technology infrastructure and abundant resources.
Many organizations now recognize the agentic era as the age of abundance, where AI presents unprecedented opportunities across every sector. But those opportunities also introduce serious operational and governance gaps that leaders need to close quickly before the competition catches up. The shift from the analytics era to the agentic era is difficult without a structured transformation plan and a strategy for retaining institutional knowledge.
As layoffs continue to increase, customer service and contact center roles have emerged as some of the hardest hit categories, with Gartner identifying generative AI and agentic AI as major drivers of contact center workforce reduction and operational automation — with engineers and coders, content writers, data entry and back-office roles, HR and payroll staff, and data analysts all following the same pattern.
Organizations Are Already Trading Labor Efficiency for Knowledge Risk
Microsoft announced a major round of layoffs in May 2025, affecting roughly 6,000 employees, reportedly the majority of them programmers, following CEO Satya Nadella’s confirmation that around 30% of the company’s code is now written by AI.
Amazon, in October 2025, announced one of the largest rounds of layoffs in its history, cutting 14,000 corporate roles as it looked to invest in AI and stated the need for a leaner organizational structure with fewer layers.
Klarna’s CEO said the company reduced its workforce by roughly 40% through AI-driven operational efficiencies and now expects its white collar workforce to shrink by another third by 2030 as AI adoption accelerates across enterprise functions.
The trend continues as organizations pursue AI-driven autonomy, transitioning humans from being the main drivers to riding in the passenger seat.
What Should Be a Top-of-Mind Priority for Leaders
As CIOs shift from the analytics era to the agentic AI era, that shift is grounded in AI’s core capabilities: faster execution and greater automation. The goals are familiar: reduce overhead costs, manage risk and compliance, and grow revenue. Across industries, a common pattern emerges as AI presents increasingly viable options to replace human labor.
But that shift brings unique challenges. What recent layoffs have in common is this: bulk replacement of the human workforce with AI agents risks losing institutional knowledge, which typically lives inside people’s heads and walks out the door the moment a seasoned employee leaves. An AI agent or model operating without that context becomes confidently wrong. Without guardrails, it can disrupt and destabilize core business operations — a phenomenon I call organizational amnesia.
Organizational amnesia is not simply about lacking good tools, capable AI models or well-managed data. It is about lacking the most critical ingredient: context intelligence.
In practical terms, context intelligence is the digital, machine-interpretable representation of how your business actually works. It means understanding customers, relationships, products, decision history, audit trails and interaction patterns. It is a shared understanding of reality, one that both AI and humans can act on, in real time, at the speed of machines.
For CIOs, context intelligence should be a top-of-mind priority. Simply having clean and centralized data is no longer enough. A structured path is needed to guide organizations from the data analytics era into the agentic era, one where AI is not just fast and automated, but genuinely grounded in how the business operates.
A Field CTO’s Perspective: What a Day with a Customer’s Data Team Taught Me About Organizational Amnesia
Recently, I had the opportunity to engage in a working session with the CIO and data leadership team at a large global travel and hospitality company, where I witnessed organizational amnesia playing out in real time.
The team was walking through their trade and group account data ecosystem. What existed was a collection of disconnected systems across their IT architecture: a legacy CRM as the aging source of truth for trade accounts, a global booking system, multiple regional CRM instances, a payment portal, a contact center interface and regional agent portals, all loosely connected through a mix of batch jobs and manual workarounds.
The room was filled with seasoned experts, and yet the deeper the discussions went, it became abundantly clear that the institutional knowledge of how their business actually worked was not captured in any system. It lived in the heads of the people sitting around that table.
One leader explained that the only way to look up a travel agent account was by phone number, a practice rooted in a time when every agency had a dedicated landline. Post-COVID, agents had shifted to cell phones, independent setups and flexible arrangements. The result was an explosion of duplicate records. If you searched by the wrong number, the system found nothing and a new account was simply created. No alert was triggered. No one noticed. The data quietly degraded over time. Now imagine deploying AI in such an ecosystem.
Another stakeholder described a payment portal that presented customers with a blank screen containing no trip information, no itinerary and no customer context. Deposits arrived, dropped into a queue and a team manually matched them to bookings. Ten minutes per interaction, on average, for a process that existed solely because the systems could not share context with each other.
When the conversation turned to why a key portion of their account data had never been migrated to their newer CRM platform, the answer was direct: the data was such a mess, and the relationships between agencies, sub-agencies, host accounts, consortia and individual agents were so layered and complex that no one had been able to configure the new system with enough confidence to make the move. In many ways, this is also a data governance failure: data needs to be defined with clear business meaning, lineage traceability, ownership and quality parameters before it can power anything reliably.
That complexity was not a technology failure. It was the accumulated, undocumented, unstructured institutional knowledge of a company that had been in business for nearly a hundred years, living inside spreadsheets, inside people’s memories and inside a legacy system the team described as being well past its prime.
What struck me most was a moment when one of the senior architects paused and said: “I want to bring it back to the data. Where is it? Where does it need to be so it can solve all of these problems?” The room went quiet. Not because the question was hard, but because everyone knew the honest answer was: we do not actually know yet.
This is organizational amnesia. It is not a technology problem. It is a context problem. The tools exist. The talent is in the room. But without a machine-interpretable representation of how the business works, who the customers are, what relationships exist and how everything connects, even the best AI system will operate confidently in the wrong direction.
The team is doing the right thing. They are slowing down to build the foundation first: defining the data model, establishing trusted master records for their account data and creating the context layer that will eventually make their AI investments pay off. That discipline is exactly what CIOs need to lead with as they move into the agentic era.
The Agentic Era Begins with a Machine-Readable View of the Enterprise
The journey from the analytics era to the agentic era is hard without a structured path to lead such a transformation. Before putting any AI system in place, leaders need to understand the context requirements and the human element behind their data. Without a proper transition plan and well-established governance processes, organizations risk confining their AI projects to experimentation that never scales, and organizational amnesia sets in.
A proper plan is not only necessary during layoffs or AI-driven workforce transitions. As organizations continue to invest more in AI and accumulate knowledge along the way, the foundations must be designed to capture context at every step, making it a shared reality for both humans and AI systems alike.
The most important question to bring to your data leadership team is this: Do we have a digital, machine-interpretable representation of our business? Do our AI systems and our people share a common understanding of who our customers are, what relationships exist, how they interact with us and where that data comes from? If the answer is not a clear yes, that is where the work begins.
