“Giving an employee an AI tool can make an existing task faster. Redesigning a workflow around what AI makes possible can change the economics and performance of the process.”
Redeployed is a weekly newsletter that breaks down one important AI story at a time for product builders and engineering leaders. Every issue explores what the shift means for technology companies and how leaders can respond.
McKinsey's latest AI research exposes a growing gap inside companies. AI adoption has reached 89% of organizations, and 80% of employees say AI improves their productivity. Yet only 37% of companies report a positive impact on EBIT, roughly unchanged from the previous year.
That gap suggests companies are getting better at making individual tasks faster without necessarily changing the economics of the business. The value of those productivity gains depends on what happens to the capacity AI creates.
Faster Tasks Don't Guarantee Better Economics
Imagine AI helps a developer complete a task in seven hours instead of ten. The company has technically saved three hours, but the financial value of those hours depends on what happens next.
If that capacity helps the team ship more product or tackle higher-value engineering work, the productivity gain can compound. If the same workflow, approval process, and delivery schedule remain in place, the company may simply have an employee who finishes one step sooner.
The same problem appears outside engineering. AI can reduce the time required to analyze data, prepare sales outreach, or create marketing assets. Each task can show a dramatic productivity improvement while revenue, margins, and overall throughput barely move.
Where the Productivity Gain Goes
Organizations are collections of connected workflows. Speeding up one part creates meaningful value only when the rest of the system can absorb the additional capacity.
A product team might produce specifications faster only to wait for engineering. Developers might generate more code while review and QA become overloaded. A marketing team can produce twice as much content without creating twice as much demand. The constraint has simply moved somewhere else.
This is where workflow design becomes important. Once AI creates additional capacity, leaders need to follow that capacity through the process and identify what prevents it from reaching a business outcome.
A recent McKinsey study offers a useful example. Software company Sonar redesigned parts of its product development lifecycle around AI agents rather than simply adding copilots to the existing process. Across three teams, McKinsey reported a 3.4x reduction in pull-request cycle duration, 2.2x higher pull-request throughput, and 50–80% productivity improvement in build activities.
The important part is what Sonar changed around the technology. Teams reworked how tasks were scoped, reviewed, tested, and handed between humans and agents. AI became part of the workflow rather than another tool sitting inside the old one.
Measure What Changed Downstream
Many companies still track AI adoption through usage: licenses activated, employees using copilots, prompts submitted, or estimated hours saved. Those metrics can show whether people are using the technology, but they reveal much less about its economic impact.
A more useful measurement starts downstream. Engineering teams can look at cycle time, reliability, and product throughput. Customer support can track resolution time and cost per ticket. Sales organizations can examine whether representatives are spending more time selling and producing more pipeline. The metric depends on the workflow, but it should connect the AI-enabled improvement to an outcome the company already cares about.
That distinction also helps separate productivity from activity. Generating more code, content, reports, or analysis has limited value if the organization cannot turn that additional output into something useful.
Redesign the Workflow Around the New Capacity
Most company processes were designed around human constraints. They contain handoffs, approval layers, staffing assumptions, and schedules that made sense when certain tasks required hours or days of manual work.
When AI changes one of those constraints, the surrounding process may need to change with it.
Consider a recurring analysis that once required two days and can now be completed in two hours. The obvious benefit is faster analysis. The larger opportunity is reconsidering how often the analysis happens, which decisions can happen sooner, and where the remaining capacity should go.
That is how a local productivity improvement starts affecting the operating model. The company can change the process around the new capability instead of simply inserting AI into the old one.
This issue of Redeployed is brought to you by Tecla Labs: As AI creates new capacity inside teams, the workflows around that capacity often need to change too. Tecla Labs helps companies identify where AI can create leverage, design the workflow around it, and build the system that makes it useful in practice. Start with a free AI Assessment to identify the workflow worth tackling first.
Turning AI Capacity Into Operating Leverage
With AI already in use across most organizations, proving that employees can use the technology is becoming less interesting. The harder question is what companies do with the capacity it creates.
A useful place to start is with workflows where AI is already producing obvious time savings. Follow those gains downstream. If one stage becomes faster, look at what happens next. Does another team become the bottleneck? Does output increase without changing revenue or cost? Does the employee simply end up with more unstructured time?
Those questions also give leaders a better way to prioritize AI investments. A modest improvement in a workflow tied directly to revenue, cost, or product delivery can matter far more than a dramatic time saving on an isolated task.
The next useful AI review may have less to do with which tools employees are using and more to do with what happened to the hours those tools saved. If nobody can trace that capacity to a shorter cycle, lower cost, better decision, or new output, the workflow probably still needs work.
Connect With Other Technology Leaders
If you want to exchange practical ideas with senior technology and product leaders navigating the AI era, check out the upcoming GILD Forums. They bring together experienced operators for peer discussions around the technology and business decisions they are working through right now.
More to come…
– Gino Ferrand, Founder @ Tecla Labs