Managing AI Adoption’s J-Curve in Real Estate

The J-curve in commercial real estate AI adoption describes an initial decline in performance before improvement: new systems disrupt established workflows, temporarily reduce employee productivity, increase support tickets and workarounds, and can leave an organization worse off for weeks or months. The depth and duration of that dip are key differences between successful and failed initiatives.

Before deploying technology, leaders need realistic expectations for what an AI initiative will accomplish and when. Chris Campbell, JLL’s Managing Director of Technology Advisory, says an early productivity dip should not be treated as proof that AI is failing; the goal is to manage the transition and shorten the time to measurable results. Promising transformative results in the first quarter can make early shortfalls look like failure. In the article’s example, a pilot promised to improve productivity by 40% but delivering 12% is shut down, while a pilot promised to improve it by 10% and delivering 12% is expanded, despite identical technology performance.

Employee engagement also affects how quickly an organization recovers. Campbell recommends starting with small, repeatable tasks that can be done faster, better, or differently, then measuring outcomes. A general claim of improved efficiency is easier to dismiss than a specific example, such as a team saving six hours a week on a task. Showing how work becomes better for employees, not only more efficient for the organization, can build enthusiasm through the transition.

Organizations can support adoption by identifying champions early across multiple functions, not only in technology teams. Campbell says champions need detailed knowledge of existing workflows, including informal workarounds and friction points, so they can redesign work rather than simply automate it. Committees that oversee AI programs also need clear priorities, such as resilience, cost optimization, and user experience, ranked to guide choices when priorities conflict.

These committees may also manage the lifecycle of AI agents, including deciding when agents should be retired. Without a retirement process, agents can continue operating without oversight, consume budget, and produce outputs that no one reviews. Campbell says agents performing business tasks need defined roles, appropriate training, clear boundaries, performance management, and a decision on whether they should continue. The article concludes that adjustment periods are unavoidable in meaningful changes to work, but realistic expectations, small projects with concrete benefits, cross-functional champions, and governance can reduce the dip and speed recovery.

The J-curve in commercial real estate AI adoption describes an initial decline in performance before improvement: new systems disrupt established workflows, temporarily reduce employee productivity, increase support tickets and workarounds, and can leave an organization worse off for weeks or months. The depth and duration of that dip are key differences between successful and failed initiatives.

Before deploying technology, leaders need realistic expectations for what an AI initiative will accomplish and when. Chris Campbell, JLL’s Managing Director of Technology Advisory, says an early productivity dip should not be treated as proof that AI is failing; the goal is to manage the transition and shorten the time to measurable results. Promising transformative results in the first quarter can make early shortfalls look like failure. In the article’s example, a pilot promised to improve productivity by 40% but delivering 12% is shut down, while a pilot promised to improve it by 10% and delivering 12% is expanded, despite identical technology performance.

Employee engagement also affects how quickly an organization recovers. Campbell recommends starting with small, repeatable tasks that can be done faster, better, or differently, then measuring outcomes. A general claim of improved efficiency is easier to dismiss than a specific example, such as a team saving six hours a week on a task. Showing how work becomes better for employees, not only more efficient for the organization, can build enthusiasm through the transition.

Organizations can support adoption by identifying champions early across multiple functions, not only in technology teams. Campbell says champions need detailed knowledge of existing workflows, including informal workarounds and friction points, so they can redesign work rather than simply automate it. Committees that oversee AI programs also need clear priorities, such as resilience, cost optimization, and user experience, ranked to guide choices when priorities conflict.

These committees may also manage the lifecycle of AI agents, including deciding when agents should be retired. Without a retirement process, agents can continue operating without oversight, consume budget, and produce outputs that no one reviews. Campbell says agents performing business tasks need defined roles, appropriate training, clear boundaries, performance management, and a decision on whether they should continue. The article concludes that adjustment periods are unavoidable in meaningful changes to work, but realistic expectations, small projects with concrete benefits, cross-functional champions, and governance can reduce the dip and speed recovery.