Inc.
AI may not need to replace knowledge workers to transform the workplace.
AI Isn’t Replacing Knowledge Workers. It’s Making Them More Powerful
Author: Ash Kumra
The popular narrative around AI in the workplace is filled with doom and gloom. Pundits constantly predict that algorithms will eliminate knowledge workers and replace corporate teams.
The reality on the ground is proving to be the exact opposite. When smart companies give their staff access to AI coding tools and teams of builders, those knowledge workers do not lose their jobs. They become more productive, transforming from passive software users into the people who shape what gets built.
For managers, the key leadership lesson is clear. The real promise of AI is not reducing headcount. It is giving the people who know the work best a way to turn their know-how into solutions for busywork, with builders alongside them to make those solutions real. The result is greater time for higher-value work.
From ideas to impact — at AI speed
Consider what happened at Three Crowns, an international disputes law firm.
Hugh Carlson, the firm’s CEO and a former software engineer, partnered with Robert Mahari, founder of the legal AI consultancy Akiva and now Head of Claude for Legal at Anthropic, to launch a program called Vanguard.
The goal was simple: show lawyers and business services staff what Claude Code makes possible, let them surface the problems worth solving. Then, pair them with a small team of builders who turn those ideas into tools.
The productivity gains quickly compounded once the team had done the reps. For example, the firm built a customized application layer that monitors global news, runs each story through a proprietary algorithm to surface only high-quality leads, screens those leads against finance and conflicts data, and generates tailored client pitches in seconds. A first-draft pitch that previously took days of manual research now happens almost instantly.
As the firm gets better at generating ideas and better at building on them, each development cycle is faster than the last.
Nobody on the business services team has been replaced. Instead, their capacity has expanded. To achieve similar productivity gains with your own knowledge workers, follow this three-step organizational framework.
1. Show your team the art of the possible.
Knowledge workers understand their daily friction points better than anyone else. However, non-technical employees rarely know what AI tools can actually accomplish.
To bridge this gap, Carlson focused on live demonstrations rather than abstract training sessions.
“First, rather than tell lawyers and business service professionals what Claude Code can do, we show them,” Carlson explained.
He noted that once employees see the tool in action, they immediately begin identifying operational bottlenecks. Staff members do not need to become full-time programmers. They simply work iteratively with technical specialists to build tailored solutions for their specific pain points.
2. Address the organizational structure, not just the technology.
Handing employees an AI subscription will not magically boost output. Traditional corporate environments are structured to route technology needs through an outside vendor or a centralized IT department. That bottleneck kills momentum.
“The biggest misconception is that this is purely a technology question,” Mahari explained.
He shared that when knowledge workers begin shaping their own tools, the traditional IT model breaks down. Organizations that successfully scale workplace productivity deliberately train employees to build responsibly, empower internal champions, establish clear deployment pipelines, and create systems to prevent duplicate work.
3. Shrink the security perimeter before building.
Speed often creates security risks. When teams outside the traditional software function start building tools that handle sensitive company data, information security teams naturally push back.
The solution is to build inside pre-approved corporate environments.
“The secret is to shrink the security surface before you write a line of code,” Mahari noted.
He explained that teams should build inside infrastructure that security departments have already reviewed. By building thin, customized application layers on top of vetted enterprise platforms, teams can safely deploy production-ready tools in weeks rather than months.
Another way to shrink that perimeter is to decide what not to build. Some Vanguard tools are deliberately small: one-off utilities that solve a single niche problem and then get out of the way. And some things should not be built at all. Foundational tools are still bought, not built.
Three Crowns has no ambition to code a proprietary version of Microsoft Word or a major legal platform, though it is making inroads into developing its own systems of record.
The aspiration behind Vanguard is to buy where fit and price both work and to build where does not.
Empower the people doing the work
The distance between the person who notices an opportunity for improvement and the person able to fix it has long been measured in budget cycles and vendor contracts, if at all. Vanguard shrinks this distance by empowering the professionals doing the work.
“The professionals closest to core business processes have always been the ones best placed to say, ‘I think we can do better here,’” Carlson said. “Now, with powerful tools like Claude Code and ChatGPT Codex, organizations can swiftly act upon these insights.”
Workplace productivity surges when you remove the friction between identifying a problem and building its solution. The companies that win the next decade will not be the ones that replace their knowledge workers, but the ones that unlock their full potential.
Credits: TCA, LLC.