The work between the work is where AI might help most.

In its 2026 State of AI survey, McKinsey found that 80% of respondents said AI had improved their individual productivity, while 37% attributed at least some earnings impact to their organization's use of AI. These are self-reported survey results, not a direct measurement of what any one company gained. The gap raises a useful question: where does the time saved by an employee go when the next step still has to wait?

McKinsey's follow-up analysis points to the handoffs between teams. A proposal may be drafted quickly but sit waiting for pricing approval. A customer request may be summarized but still need someone to find the contract, check what was promised, and route the issue to the right person. An invoice may be prepared in minutes and then spend days in a back-and-forth over missing information. Faster drafting does little if the process around it stays the same.

This does not require a company to automate every decision or remove the people in the middle. The first step is to find out what those people are actually doing. Are they checking facts, applying judgment, reconciling two systems, or waiting because nobody knows who owns the next step? Some checks should stay human. Others could be supported by AI or a clearer workflow. The distinction only becomes visible when someone maps the entire path, including the time between tasks.

I think this is also why an organization needs to be on the same page about adoption. If each employee experiments alone, nobody is responsible for the whole process. A useful roadmap names the workflows the company wants to improve, puts them in an order tied to business priorities, and gives each one an owner. For each workflow, the team should agree on the inputs, the context AI needs, what a good result looks like, and where human review belongs. It should be clear who can approve a change and how the team will know whether the new way of working is helping.

This is where Ruby's discovery work starts. We sit with the people involved, document the repeated questions and handoffs, and build that roadmap with the team. A workflow should connect to something the business already cares about, such as a faster client response or fewer corrections before a report goes out. We identify the person who owns it, test a change with the employees who do the work, set a clear review point, and measure the full cycle rather than only how long the first draft takes. As the tools change, the roadmap can change with them.