_{Whoa, that’s deep}

One of my favorite moments in the workshops I run happens early on, when I ask people who have used “Deep Research”. I scan the room — maybe thirty transit professionals — and three hands go up (most people think “Deep Research” is a ChatGPT competitor). Then I ask the follow-up: "What scares you most about AI?" and suddenly every hand is in the air. Hallucinations. How do we trust the information it's giving us?

I see those three hands and those thirty hands and I think... nice. Because I'm about to show them the answer to both questions at once.

Deep Research is like a Google search on steroids. It actually wants you to understand the answers, not just find them. It's like asking your most diligent senior director to go research everything they could possibly find about a topic — every TCRP study, every academically published paper, every vendor website, every case study, every board report that could possibly be known to man — and then getting the results in 10 minutes instead of weeks, complete with all the citations you'd need to verify it yourself. It's available in every frontier AI model, including Gemini, ChatGPT, Claude, and yes, even CoPilot, usually under a dropdown of some sort. Some require a paid subscription to use Deep Research, some offer a handful of Researches for free.

Let me give you three examples of how I actually use this.

As I'm building my team at KTS, I was trying to understand what salary range would be competitive for a role I was hiring. I gave my job description to Claude's Deep Research and asked it to find competitive salary ranges for this kind of position in my geography. It came back with real data and it actually encouraged me to increase the rate I was planning to offer. Instead of guessing, I was able to see the landscape and make a confident decision backed by evidence. You can run that same wage check search for your operators! In five minutes, you'd know whether your wages are competitive or whether you're losing good people to Amazon because you're $2.50 an hour behind and didn't know it.

Or consider this non-transit example. My girlfriend is a jeweler who was recently invited by Vogue Magazine to display her work at the Four Seasons. It was a sizable investment, and she wanted to know what the expected return would be. We used Deep Research to understand all past vendors who had attended similar events and track the growth of their brands using publicly available sales data, so we could get a better estimate than just our gut on whether now was the right time for this marketing investment. This wasn't a business decision I had any expertise in — but Deep Research gave us enough of a foundation to make a real decision with confidence instead of hope.

And then there's the example that surprised me most. I wanted to know the total cost of acquisition for a large transit agency's recent software procurement from three years ago. Previously, I would have Googled, clicked around, dug through board reports, tried to find the right dates, Control-F'd the vendor name in 100-page board PDFs — you know the drill. It would have taken forever. Instead, I ran Deep Research and got a surprising answer: there was simply no publicly available information on this. When Gemini told me this, I ran a Deep Research with Claude instead. But when both confirmed it, I was compelled to trust that this information genuinely didn't exist in the public domain. Deep Research saved me hours of fruitlessly clicking around. And remember those thirty hands worried about hallucinations? This is how you build trust — when a tool tells you "I looked everywhere and it's not there" instead of making something up, and when you can verify that by asking a second tool the same question. 

Here's the ultimate unlock for me, and it goes beyond any single example. Deep Research isn't just a search engine — it's an educator. It teaches me everything I didn't know that I didn't know about a new topic. In a recent workshop, I used Deep Research to identify two meaningful grants for a fictional transit agency in the Northeast, and it didn't just find the grants — it taught us what format the application needed, what made past winners successful, and what made others fall short. That's the shift. It doesn't just find information; it builds your understanding from the ground up. 

This is particularly valuable for small transit agencies considering a big once-in-a-career move like introducing microtransit, building a new facility, or becoming a Large Urban System. These are the moments where you simply don't know what you don't know, and Deep Research can take you from understanding 10% of the landscape to 70% in the time it takes to make a coffee. The cost implications are huge, because consulting contracts often charge tens of thousands of dollars just to synthesize information that was already available. Deep Research can do that in minutes, so that when it is time to bring in a consultant to manage the art of transit, you can do so cost-effectively and under a scope that is clear and educated. Because now you can know what you don't know. 

The internet democratized information. Through Deep Research, AI is democratizing understanding.

Now go make a coffee and learn something.

TransitGPT is where I share practical AI insights for transit leaders. Subscribe if this resonates or forward to a colleague who needs it. And hit reply — I read everything.

Go shatter your glass ceiling.