Seth Gilford and Steve Leathers are commercial real estate brokers in Austin. They specialize in medical office buildings — the kind of property that serves doctors, dentists, and healthcare groups.
When I sat down with them for a working session, they handed me a prospect spreadsheet. Hundreds of rows. Names, companies, properties, phone numbers.
The ask: can we personalize cold outreach for all of these?
The old way to do this is painful. You pick a name, look them up, find their LinkedIn, read about their company, figure out why your product or service is relevant to their specific situation, write a message that references something real, move to the next person, repeat for hours.
That’s why most cold outreach isn’t personalized. It’s not because people don’t care. It’s because real personalization, at scale, is just not humanly feasible.
We ran a Lindy agent instead.
What the Agent Did
The setup was straightforward. The agent connected to the spreadsheet, ran Perplexity research on each person and company, and generated a personalized LinkedIn or email message for each row. The whole batch took under 30 minutes.
But then Seth noticed something.
“How did it know this building was physician-owned?”
He was pointing at one of the messages. The draft referenced that the building appeared to be physician-owned and framed the outreach around a potential sale-leaseback opportunity.
Nobody had told the agent any of that. The spreadsheet didn’t specify ownership type. We hadn’t written any instructions about sale-leasebacks or medical office specifics. There was no if-then rule for “if physician-owned, then mention sale-leaseback.”
The AI had found it through research. It looked up the building, found information suggesting physician ownership, understood what that means in a commercial real estate context, and surfaced a relevant angle — without being asked.
Seth’s reaction was the one I watch for in every demo: “We didn’t even specify that the list had physician owners and real estate owners. We just gave you a list.”
The Difference Between Execution and Judgment
Most people think of AI in terms of execution. Give it a task, it does the task. Write this email. Fill in this template. Summarize this document. Faster, cheaper, same output.
That framing is accurate for a lot of use cases. And it’s genuinely valuable — automating execution can save enormous amounts of time.
But the physician-owned discovery isn’t execution. Nobody specified the task. The agent encountered information, interpreted what it meant in context, and decided to surface it as relevant.
That’s closer to what a good research analyst or an experienced junior colleague would do. You hand them a list, they come back not just with completed tasks but with observations: “By the way, I noticed this one seems to be a physician-owned building. You might want to mention sale-leaseback potential.”
This is what the “agent as teammate” idea actually looks like in practice — an agent that doesn’t just complete tasks but brings context and judgment to the work.
Why This Changes What’s Possible in Outreach
The conventional wisdom about personalized outreach is that it doesn’t scale. You can send a thousand generic emails, or you can send ten genuinely personal ones. That’s the trade-off.
AI breaks that trade-off.
I built a similar prototype for another client earlier in the year. The system took a prospect list, ran AI-powered research on each individual, and generated personalized LinkedIn messages — including specific details about their company, their recent work, and why the outreach might be relevant. What would have taken hours of manual research per contact ran in minutes for a full list.
The insight that sticks with me is this: the bottleneck in most sales isn’t the send. Everyone can send an email. The bottleneck is the research — figuring out enough about each person that the message actually feels personal and relevant.
When AI handles the research, the bottleneck disappears. And you get something else too: the AI sometimes surfaces angles you wouldn’t have thought to include.
What This Looks Like in Practice
For the real estate brokers, the workflow is now roughly:
- Update the prospect spreadsheet with new contacts
- Run the agent
- Review the drafted messages and tweak any that need adjustment
- Send
The research that used to be hours per person now runs in batch. The outreach is more personalized than they could have done manually. And occasionally, the agent notices something — like physician ownership and sale-leaseback potential — that adds a genuinely useful angle.
This isn’t magic. It’s an agent with access to a research tool, given instructions about what kind of insights are relevant for their business. But the output is qualitatively different from what the brokers were producing before.
Where to Start
If you do any kind of outreach — sales, partnerships, hiring, networking — the question is worth asking: what research is the bottleneck for you?
Is it understanding each company before a first call? Is it finding the right angle for a cold email? Is it knowing who to prioritize on a list?
Start there. Describe what a great junior researcher would do for you before each outreach. That description is your agent spec.
If you want to see how this kind of agent gets built from scratch, the 4-Day AI Sprint walks through prospect research, personalized outreach, and other practical workflows step by step.
Recommended for you
Want the full system? 25X is the flagship productivity system we teach.
Explore 25X →