What's the best use case for generative AI?

By Jamie Biesiada | Mar 24, 2025

WASHINGTON — Client communication is a pretty obvious area of a travel advisor's business where generative AI could be useful. After all, it can write pretty much anything, from emails to blog posts and beyond. But technologist Nilesh Ashra doesn't believe that is the best use case for GenAI.

"Could we have this communicate with our clients? Yeah," Ashra said. "Should we choose to do that? I don't think so. I really don't think so. I think that would be trading down on the emotional value of why we exist in the first place."

Ashra is the founder and CEO of the companies OK Tomorrow and Pragmatic Futurism. He spoke and offered classes at the recent Advisors in Bloom conference from Montecito Village Travel held at the Park Hyatt here.

"Generative AI like ChatGPT isn't overhyped, it's underhyped, and some business leaders have been slow to catch on," he said onstage.

With every job comes administrative work and creative work, Ashra said. While tech companies have sold GenAI as a solution to administrative work, he believes its real power comes in helping workers with their creative work.

"I don't need help writing the damn email, I need help with the things inside it," he said. "It's not about the emails, it's about the ideas."

A use case for generative AI

Ashra uses generative AI to clear his mind when he's stressed at work, for instance. He activates voice mode in ChatGPT with AirPods in, then takes a walk. He prompts the AI to listen to everything he says, and whenever he pauses, ask if anything else is on his mind.

When the walk is done, he has a transcript he can ask ChatGPT about. Are there any patterns? Can it use the information within to make to-do lists, one of things that take less than 5 minutes to complete and one of things that take longer? 

During a workshop, he suggested advisors use a transcription service like Tactiq or Otter to record client conversations. The transcripts can then be uploaded into a generative AI service and mined for information, like a client's biggest concerns or questions to ask the client to better understand them.

Multiple conversations over months could also be uploaded for the same purpose, as could meetings with suppliers. 

Ashra also provided an example of how to use generative AI for marketing purposes. He pulled a McKinsey report on the state of tourism and travel and uploaded it, then asked the AI to summarize the 68-page document. A step further could be asking the AI to pull information included on wellness in the report for a marketing campaign, then create a table with key themes and marketing messages that could speak to them.

You get out what you put in

One of the most important things to keep in mind with GenAI is the context window, Ashra said. That essentially encompasses everything a user has input into a chat with generative AI. The amount of text that can be added to the context window has grown exponentially in just a few years.

Ashra recommends spending time on what is entered into the context window. A good prompt might take 15 to 20 minutes to write.

Users should give AI their goals, who they are and the role they want it to play. They should be specific about the return format (like telling the system not to use hyperbole and not to sell or persuade on anything). Then, offer caveats, like avoiding general responses and encouraging the AI to ask for clarification or input. 

Ashra also wants people to stop using generative AI as Google.

It's not good at getting facts right, he said. As an example, he asked ChatGPT if 1,000 was bigger than 1,062 — it answered affirmatively. He also asked how many Rs were in the word "strawberry," to which ChatGPT confidently answered two. When he pointed out that wasn't correct, it agreed, and said there was one.

"This is not a criticism of it," he said. "This is to focus on what it's really for."

CONTINUE READING THE ARTICLE HERE