Jen Taylor started using AI like many converts. It was when I suddenly felt that the fluency of other people’s AI was not good enough for my own. She was having dinner with her cousin, who casually mentioned that he uses AI every day at work. Taylor asked her to show him how. Just watching the conversation unfold on my laptop screen was enough. She went home and decided to use it every day, even in situations where she didn’t think it would work, to see what it could actually do.
That instinct to explore before optimizing now shapes how she helps arts and culture organizations implement AI in her role as Director of AI Strategy and Integration at Capacity Interactive, an agency that works exclusively with arts and culture clients across the United States. Taylor began her career marketing Off-Broadway and Broadway productions, spending nearly 12 years at A&E Networks, leading digital audience growth across ad-supported and subscription streaming, and found a way to integrate both threads in what she calls close to her dream job.
This episode is part of Branch’s How I Grew This series, distributed through Business of Apps.
Where most teams get stuck
When Capacity Interactive surveyed arts administrators before building its AI service, the results were more promising than expected. People were curious and generally more excited than skeptical. However, two concerns consistently surfaced: whether AI aligns with the organization’s values and mission, and how to actually learn how to use AI. Education and training are ranked as the main barriers to adoption. The findings that Taylor points to appear in most research on the subject, far beyond the field of art.
The underlying problem is primarily conceptual, not technical. Tools with a defined single purpose can be easily placed within a workflow. AI resists that framework. It’s very flexible because it can support a huge range of tasks, but it can make it difficult for teams to know where to start. Taylor uses the analogy of a Swiss Army knife, where you don’t know what half of the tools will be used for. The tools are there. Using them successfully requires a different kind of direction.
Her approach at Capacity Interactive follows a deliberate sequence. Before discussing use cases, organizations need perspective on what is and isn’t acceptable, what platforms are teams allowed to use, and where the boundaries are. Its internal policy comes first. Training continues on how to effectively prompt, manipulate image and video generation as needed, build custom tools, and more. Only then does the conversation move to specific applications tied to business outcomes. Teams that skip the first two stages and move straight to use cases tend to have shallow adoption that doesn’t stick.
How AI is changing creative teams
Source: App Business via YouTube
prompting gap
One of the most common types of failures Taylor has observed is weak prompts and disappointment with the results. The conclusion people draw is that AI doesn’t work or won’t work for them. The real issue is usually the quality of the interaction, not the functionality of the model.
A good prompt means giving the tool enough context to work with, including audience, purpose, constraints, and tone. It means treating interactions as conversations. Taylor’s own early attempts, she explains, were no better than typing “blue sweater” into the search bar and being surprised by the results, which missed the point. While this improvement has come from learning how to better structure questions, practical skills gaps remain for most teams using these tools.
From workflow to strategy
The AI efficiency case is the one that most teams arrive at first. I have a task that is running repeatedly. AI can do it faster. This is a real benefit and critical for organizations with limited resources. But Taylor’s more interesting discussion is about what’s upstream in execution.
Someone recently told her a term that she found helpful. This means that you can plan and strategize. Most teams are under pressure to consistently run campaigns and stick to a plan. AI creates room for a shift in strategy. That means pressure testing your gist, exploring how your message is likely to be received by a particular audience, and identifying where your plan has weaknesses before you take it to market. When a team is working at full capacity, that kind of thinking tends to take priority. AI cannot replace thinking, but it can make thinking more accessible.
A concrete example of her current work shows how this could play out. Capacity Interactive’s content team operates on the belief that content has become the primary targeting mechanism, backed by Meta’s proprietary guidance. Rather than writing copy for a single, defined audience and relying on precise targeting to reach them, this approach is to create messages that speak to multiple audiences and have those most likely to engage see your message on the platform. Taylor built an internal tool that takes a set of ads and a set of target audiences, reviews copy against both, identifies additional audiences that the messaging is likely to reach, and outputs draft copy for those additional audiences. This tool is not intended to make strategic decisions. This expands your team’s ability to act on the strategic positions they already hold.
Consistency as an underrated use case
Symphony’s work, where multiple departments independently create external communications, has resulted in a different type of application. The messages were inconsistent and difficult to manage through traditional editorial processes. Taylor’s team built a tool trained on the organization’s brand voice and style guide. Any department can now run copy to check their written content or generate a first draft for review before publication. This tool does not produce final output. Standards are held while humans decide what to say.
This shows what Taylor reiterates throughout the conversation. The value of AI is not in removing human judgment from the process, but in supporting it more consistently. The risk she sees in a fully automated content pipeline—one in which sentiment is captured from a website, copy is written, and posts are published without anyone reviewing the output—is not primarily that things can go wrong in a tangible way. That means the work will be common, difficult to detect, and difficult to undo.
Measurement and insight issues
When it comes to measurement, Taylor believes that AI won’t change what good measurement looks like. The principles remain the same. What changes is the ability to uncover patterns in your data that would otherwise require significant manual effort to find.
Organizations that previously had access to data but lacked the ability to regularly extract meaningful insights from it can benefit from tools that can do that work more quickly. Output still requires human validation. The model that identifies an insight does not make that insight accurate. Taylor is adamant that those responsible for the work have a responsibility to see what a tool will produce before acting on it.
AI as infrastructure
The longer part of Taylor’s argument is that current planned AI strategies will eventually be replaced by things like basic infrastructure. The Internet has undergone a similar transition. From clear strategic questions that organizations must answer to assumed conditions for how work should be done. AI is moving in the same direction, and teams that treat AI as a permanent experiment rather than an evolving capability may end up behind the transition rather than ahead of it.
What that infrastructure will actually look like is still taking shape. Taylor points to two directions she’s looking at. The first is a more connected system. This is a tool that gives you access to more context about your business and can produce more relevant output without requiring you to provide all the background each time. The second is a proactive surface. Tools provide users with relevant information rather than waiting for questions. Both represent a shift from AI as something that goes to AI as something that exists within the work itself.
For now, her practical advice for anyone using these tools is simple: Stay informed. Use AI to augment your abilities, not to replace your judgment. And learn how to prompt well, because the quality of what you get depends on what you spend more than most people appreciate.
