Key Takeaways
- ATMs never killed bank tellers; the smartphone did. Teller headcounts actually grew after automated teller machines launched in the 1970s, only collapsing decades later when mobile phones reorganized retail banking.
- Surface-level wrappers hide the mechanics of automation. Borthwick argues operators must look past apps like Instinct or Muse to test agentic engines like OpenClaw and Hermes directly.
- David Weisburd points to a structural shift in software development: “For the first time in history it's never happened before. Coding is in the English language.”
- Managing software agents requires disciplined feedback loops. As Borthwick explains: “Whenever I get with my agent and go through a difficult set of tasks, I will then afterwards invariably ask my agent to do a review. So I'll say my agent, okay, what did we learn from that?”
- Defending white-collar margins requires executing Borthwick's 3-Step Strategy for White-Collar AI Antifragility to broaden individual operational scope.
The Borthwick's 3-Step Strategy for White-Collar AI Antifragility
Step 1: Deeply Understand and Engage with the Technology
Do not just take packaged, off-the-shelf consumer apps like Instinct or Muse. Move past the surface package and experiment directly with underlying agentic harnesses such as OpenClaw or Hermes to understand the layers of the stack and how agentic workflows operate.
Step 2: Unbundle Your Existing Job Responsibilities
Examine the bundle of tasks and workflows that define your role today. Systematically audit which specific components can be performed, streamlined, or automated using AI tools.
Step 3: Expand the Surface Area of Your Work
Use AI to operate across a broader surface area that was previously inaccessible due to resource or time constraints. For example, an entry-level paralegal can use AI to instantly evaluate every precedent case in firm history and the broader legal universe, fundamentally redefining the output and scope of their job.
When This Works (and When It Doesn't)
This framework applies to knowledge workers in fragmented, information-dense sectors such as corporate legal diligence, market research, and pre-seed venture sourcing. In these environments, professionals face clear performance caps imposed by human processing speed. When an associate can deploy autonomous agents to parse ten thousand contract filings instead of fifty, their value expands. They stop operating as manual information filters and start operating as system coordinators who govern volume.
It fails in closed, compliance-heavy corporate bureaucracies with strict data air-gaps and static compensation bands. If a mid-level analyst in an inflexible back office automates forty hours of spreadsheet reconciliation down to five minutes, the institution rarely expands their scope. Instead, management views the efficiency gain as headcount redundancy and eliminates the seat. Expanding your operational boundary only protects your position if the firm has the structural appetite to capture that excess throughput.
Why It Matters
Borthwick's thesis signals how early-stage tech investing and private equity operating groups will assess human capital efficiency. The historical teller precedent exposes why headline panic about rapid white-collar layoffs misses the real mechanics of displacement. Legacy staff counts do not evaporate overnight when a new efficiency tool appears; they hold steady until a different platform redesigns the distribution layer.
For sponsors backing service-heavy portfolio companies, multiple expansion will no longer come from labor arbitrage or offshore headcount. It will track the ratio of software steerers to raw output. As natural language replaces procedural code, enterprise software pricing power will migrate away from seat-based licenses toward autonomous workflow execution. Value will flow to businesses that price on processed volume rather than billable hours.