AI Made the Work Faster. So Why Didn’t the Work Get Faster?
Companies are investing heavily in AI because they expect it to improve productivity. Oracle recently shared an example of what can happen when that expectation is actually met—and why leaders need to look beyond the productivity gain.
According to reporting from an internal Oracle town hall, the company rolled out ChatGPT Enterprise and OpenAI Codex broadly in April and May. Within three months, Oracle reported about 80% adoption. The impact on some software development work was significant. Oracle’s CIO said developers could generate code in about a week that previously might have taken a team two or three quarters to develop.
That is a substantial improvement in productivity. But producing code is only one part of getting software to customers. The code still has to move through testing, validation, deployment and release processes, and those parts of the work weren't moving at the same speed. Oracle's co-CEO said the company is now working to redesign those processes to keep up.
This is where the story gets interesting from a workplace performance perspective. The AI didn't fail. It appears to have made employees significantly faster at part of their work. Tool Support improved, and that improvement exposed a Shared Capacity problem somewhere else in the process.
When Better Tools Change the Work
Tool Support is one of the six JL³ Performance Levers. It looks at whether employees have the tools and processes they need to perform successfully. A tool that allows someone to complete in a week what previously required months is a significant change in that workplace condition.
But performance doesn't depend on one condition operating independently. Changes in one part of the work affect the demands placed on other parts.
That's where Shared Capacity comes in. Shared Capacity looks at whether the people, time and organizational resources available are sufficient for the amount of work that needs to be completed.
Consider what happens if a team normally produces 20 pieces of work each week and another team is responsible for reviewing those 20. If a new tool suddenly allows the first team to produce 60, the organization has increased production without necessarily increasing its ability to handle what is being produced.
The first team's productivity numbers might look great. The final result might not improve at all.
The Delay Didn't Disappear
This is an easy problem for leaders to miss because the new tool can be doing exactly what it was purchased to do.
Employees are completing tasks faster. Output is increasing. The organization can point to measurable productivity gains.
Meanwhile, work may be accumulating somewhere else.
That doesn't mean the technology investment was a mistake. It means the change affected more than the employees using the technology. Leaders have to examine what happens before and after the work that changed.
If AI allows a salesperson to generate significantly more proposals, can the people responsible for reviewing and approving them keep up? If automation allows a company to process customer requests faster, can the team responsible for fulfilling those requests handle the additional volume? If one department becomes substantially more efficient, what happens to the departments that receive its work?
Those are capacity questions created by an improvement in Tool Support.
Look at the Whole Process
The Oracle example is useful because many organizations are currently asking how much productivity they can gain from AI. That's a reasonable question, but it shouldn't be the only one.
Leaders should also ask what happens to the work after AI makes part of it faster. Where does the work go next? Does that part of the organization have enough capacity to handle the change? Has work actually been eliminated, or has some of it moved to another person or department?
Most importantly, did the organization's final result improve?
Oracle's experience shows why those questions matter. Its executives described dramatic improvements in coding speed while also acknowledging that those improvements had not yet translated into getting products to customers faster.
That's an important distinction. Making a task faster is not necessarily the same as improving performance.
Sometimes a new tool solves exactly the problem it was intended to solve and reveals the next workplace condition that needs attention.
What Is Making Performance Harder for Your Team?
When performance isn't improving as expected, the answer isn't always another tool or more effort from employees. The conditions surrounding the work may be limiting what those improvements can accomplish.
The JL³ Performance Pulse™ helps leaders examine six workplace conditions that can make performance easier or harder, including Tool Support and Shared Capacity.
Take the JL³ Performance Pulse™ to identify which workplace conditions may be affecting your team's performance.