Marc Benioff has been open about it. Salesforce reduced its customer support workforce from around 9,000 employees to approximately 5,000, deploying AI tools to handle work that had previously required humans. The company has framed this as an operational success story — AI now handles roughly half of support workload — and the underlying logic is consistent with how most modern companies are thinking about the opportunity.
There's a part of the story that hasn't been discussed publicly, and it matters more than the headcount number. When 4,000 experienced support professionals leave a company, something leaves with them that is not captured in any operational metric. It's the 18 months of learned context about a specific enterprise customer. It's the tacit knowledge of which product quirks matter to which verticals. It's the judgement about when to escalate and when to resolve directly. It's the relationships.
That body of knowledge is an intangible asset. It's called organisational capital, and almost nobody measured what happened to it.
Organisational capital is the intangible most likely to be lost during an AI-driven workforce transformation and the least likely to appear on any management dashboard. When it disappears, you rarely see the cost until customer satisfaction and operational reliability begin to drift — usually three to six quarters after the change.
Organisational capital is one of the twelve drivers in the Opagio 12 framework. It's also one of the most systematically undervalued. It covers the accumulated know-how, processes, playbooks, documentation, and institutional memory that make a business work. In academic economics it sits alongside R&D and software as a core category of intangible capital — the economist Carol Corrado and her co-authors have spent two decades documenting how large a share of modern GDP growth comes from this asset class.
The challenge with organisational capital is that it is distributed across people, systems, and informal networks. When a senior support engineer leaves, they take with them:
An AI agent can be trained on documented processes, but it cannot be trained on what was never documented. The knowledge that exists in heads is the knowledge most at risk during a workforce transformation.
Josh Bersin's recent analysis of the AI workforce story made a related point. His research across seventy companies found that organisations treating AI as a tool for individual productivity gain rarely achieved meaningful job reduction. The transformations that did produce material workforce changes required deep process re-engineering — which in turn required a deliberate capture and redesign of institutional knowledge. The companies that got this right invested in knowledge transfer before layoffs, documented critical workflows, and structured the AI deployment to absorb what the departing workforce knew. The companies that didn't found themselves retraining the AI on problems their departed staff had already solved.
The measurement implication is specific. A company executing an AI-driven workforce transformation should be tracking organisational capital as a dedicated metric — before the transformation, during it, and after. The question isn't whether the remaining team can handle the workload with AI assistance. The question is whether the institutional knowledge base has been preserved, transferred, or rebuilt in a form that supports ongoing operations.
This is not a theoretical exercise. The practical measurement covers several dimensions.
| Dimension | What to measure | Why it matters |
|---|---|---|
| Documentation density | What percentage of critical workflows are captured in a form that an AI can use? | The raw material for automation only exists where the process has been written down. |
| Process redundancy | If a critical process currently depends on a single individual, what's the plan for that dependency? | Single-person dependencies become single-person failures in a transformation. |
| Knowledge transfer completeness | Of the leavers, how many have participated in structured knowledge capture before departure? | The window to extract tacit knowledge closes when the person walks out. |
| Error recovery capability | When the AI encounters a situation outside its training, what's the escalation path to human expertise, and how deep is that remaining bench? | The remaining senior bench is the safety net. Thin it too far and the reliability drops with no early warning. |
The companies that get this right produce a specific kind of dashboard. It shows organisational capital trend lines through the transformation. It flags high-dependency processes that haven't been documented. It tracks the knowledge transfer completion rate per team. It correlates customer satisfaction changes with organisational capital movements. In short, it treats the institutional knowledge base the same way a CFO treats working capital — as a measurable, managed asset.
A mid-cap SaaS company reduces its support team by 35% over six months and deploys an AI support layer. In the transformation plan, each departing engineer participates in a structured knowledge-capture exit — two weeks of documented case walkthroughs, tagged by customer and product area. The documentation feeds the AI's retrieval layer directly. Six months after the cut, the organisational capital index is down only 8% against a 35% headcount drop. The equivalent competitor that did not structure the exit is down 32% on the same index. Both companies look similar on the P&L. The gap shows in customer retention two years later.
When we built the Opagio 12 framework, organisational capital was one of the harder drivers to operationalise. It doesn't have a natural financial line item in the way that R&D spending or marketing spend maps to the technology and brand drivers. The methodology covers it by looking at a combination of process documentation maturity, dependency concentration, knowledge transfer activity, and the operational performance indicators that correlate with organisational capital health — support quality, error rates, resolution times, employee tenure, and retention of critical roles.
For a company going through the kind of transformation Salesforce has executed, the measurement view is diagnostic. It shows whether the drop in Human Capital has been offset by an equivalent strengthening of Technology and Organisational Capital, or whether the organisation has quietly moved into a weaker intangible position that will compound over subsequent quarters. The companies that see the latter early have options. The companies that only see it when customer satisfaction numbers shift have fewer.
This pattern is one of five drivers that typically move during an AI workforce transformation. For the full picture, see Block's 40% workforce cut on the portfolio trade-off, and the seven-driver CFO breakdown on what most AI investment cases miss.
The Salesforce transformation will be evaluated over years, not quarters. The company may well come through it with a stronger intangible base than it started with — Benioff has signalled a deliberate operating model shift, not a short-term cost cut. The critical question is whether the internal measurement is granular enough to show, driver by driver, where the transformation has created value and where it has destroyed it.
The same question applies to every company following this playbook. Institutional knowledge is the intangible asset most likely to be lost during an AI-driven transformation and the one least likely to show up in the monthly management pack. It deserves its own line on the dashboard.
If your leadership team is about to shrink a customer-facing or knowledge-intensive function, the first move is to know what you have now. Two ways to start:
The knowledge inside the heads of the team you are about to restructure is one of the most valuable assets your company owns. It deserves the same treatment as working capital.
Block cut 40% of its workforce and tied it directly to AI. The market cheered. But a workforce transformation at this scale moves at least five intangible value drivers simultaneously — most companies measure two. Here's what that gap actually costs.
Read more →
Every technology wave builds its own measurement discipline. Manufacturing gave us cost accounting. Programmatic marketing gave us attribution. Subscription software gave us ARR. AI is in the gap now — and 90% of firms report zero measurable productivity impact from their AI spending. Here's what the missing measurement layer needs to do.
Read more →
How blockchain technology and smart contracts are classified as technology-based intangible assets under IFRS 3. Covers the distinction between blockchain infrastructure, protocol-level IP, tokens, and smart contract platforms.
Read more →Get the latest insights on intangible asset growth and productivity delivered to your inbox.
Book a free consultation to see how Opagio Intangibles can help your business.