AI emerged in nearly half of Flow Digital’s client conversations. Yet, actual intent was lower: only 20% of organizations explicitly asked for help implementing a specific AI use case. Among this active group, 22% have progressed to a live deployment or an active build.
The unexpected finding is what happened when clients didn’t ask for AI. Nearly half (47%) of the AI implementations Flow Digital delivered were for clients who had initially described an operational problem rather than requested an AI solution.
The results are from Flow Digital’s original research, The AI Use Gap, which examines six months of recorded sales and client calls, alongside implementation work, from early to mid-2026. The sample covers hundreds of organizations across several industries.
The figures follow distinct stages of the process. AI appeared in 49% of conversations, including occasions when Flow Digital raised the topic, it was mentioned in passing, or the client discussed an existing tool with AI features. The 20% figure refers to organizations that independently requested help with a specific AI use case. Within that group, 18% had a solution deployed, and another 4% had one actively being built. The remaining 78% had not progressed to a build during the period analysed.
Flow Digital’s work also revealed why an initial AI request might change. In one documented project, the team replaced an AI-based matching step with conventional code because the matching rules were stable. In another, an AI calculation failed in testing and was replaced with a formula. A decision not to build an AI component can be the outcome of examining the workflow, rather than a stalled project.
Meanwhile, clients who did not request AI sometimes described bottlenecks that called for it: emails arriving in different formats, information trapped in documents or requests that staff had to categorize by hand. Those tasks can require interpretation before conventional automation updates a record, routes the work, or sends an uncertain case to a person.
Among Flow Digital’s live and in-progress AI work, 49% focused on extracting information from documents or emails. Of the AI components built or actively being implemented, 61% operated as steps inside Zapier workflows. Much of the AI that reached production was therefore a small part of a wider process, rather than a standalone assistant or agent.
As Nathan Weill, Flow Digital’s founder and the author of the research, writes, sometimes the most valuable recommendation is: “You do not need AI for this.” The analysis suggests that the question for businesses is less whether they can add AI to a process, and more which step, if any, actually benefits from it. Flow Digital applies that approach in its AI automation services.
About the Research
Flow Digital analysed recorded sales and client-call transcripts from a six-month period in early to mid-2026, alongside its implementation work. The sample included hundreds of organizations across several industries. Findings are measured at the organization level unless otherwise stated, and percentages are rounded.
About Flow Digital
Flow Digital is a workflow and business automation consultancy that helps organizations improve processes, connect software, and implement automation, including AI where it fits the task. Its AI automation work combines process design, software integration, and implementation.
###
Contact
Flow Digital
Caitlin Williams
300 Kent Ave Ste 902, Brooklyn, NY 11249
(617) 594-9351
https://flow.digital/
LinkedIn: https://www.linkedin.com/in/nweill/
Source: 38 Digital Market News
Release ID: 2747056
