DEV Community

Cover image for I Taught My AI Agent to Find What Isn't There. It Found Million-Dollar Clients.

I Taught My AI Agent to Find What Isn't There. It Found Million-Dollar Clients.

As developers, our work often revolves around finding things: a record in a database, a specific string, a file in a directory. But I recently ran into a much more interesting business problem: how do you use AI to find something by its absence?

The specific challenge was to find high-net-worth wealth management firms (managing over $100M) that lacked a modern, secure client portal.

This is a tricky problem. You can't just grep for "no client portal." It's a negative space search. An absence of evidence isn't evidence of absence, unless you're very clever about it.

The Agent's Logic

I built an AI agent using Python and Ollama with a two-stage logic:

  1. Positive Signal Identification: First, the agent scans the web to find firms that explicitly match the positive criteria: they are independent RIAs, they publicly talk about their "high-touch, personalized service," and they manage over $100M in assets. This creates a pool of potential targets.
  2. Negative Signal Verification: Next, for each target in the pool, the agent performs a deep crawl of their website. It actively searches for keywords like "client login," "portal," "sign in," or any <iframe> that might embed a third-party portal.

The core of the logic is a simple but powerful conditional:

if (firm.matches_positive_signals and firm.has_no_login_portal):
    flag_as_high_value_lead()
Enter fullscreen mode Exit fullscreen mode

This process effectively filtered a massive, noisy industry down to a tiny handful of "whales"—highly valuable firms with a clear, verifiable technology gap.

The Unexpected Insight: Finding the "Backdoor"

The agent found a few of these direct "whale" leads. But then it found something even more valuable: a company called selectadvisorsinstitute.com.

This company isn't an RIA itself. It's a B2B service that trains and consults for hundreds of other RIAs.

This is the strategic backdoor. A partnership with this one company is a gateway to their entire client base. It's a "one-to-many" sales opportunity, which is infinitely more scalable. My agent, by following the data, had uncovered a business strategy, not just a list of names.

The Result

This experiment proved to be incredibly effective. Finding these high-value gaps is a repeatable process, but it's time-consuming to run. I've packaged the results of this wealth management run—the "whales" and the strategic "backdoor" partner—into a single executive briefing.

It's a testament to the power of using AI not just for simple queries, but for sophisticated, logic-based discovery.


You can see the full output of the agent in my executive briefing on Gumroad: https://agenticsales.gumroad.com/l/wealth-management-leads

Top comments (0)