Tulsa Tech Week recap

I Built an AI to Find Your Customers. Now What?

Practical takeaways from Tulsa Tech Week, plus a few ways to start putting AI to work.

At Tulsa Tech Week, Patrick Acosta, SVP of Audience at Aletheia Marketing & Media, shared what happened when he set out to build an AI-powered audience intelligence platform from the ground up.

Request Patrick’s presentation Go a step further

The short version?

AI can do some absolutely remarkable things. It can also be remarkably confident when it’s wrong.

Patrick demonstrated how AI helped him compress work that traditionally required multiple people and weeks of analysis and programming into a dramatically faster process. He also shared the other side of the experience: endless iterations, unexpected errors and AI occasionally “helping” by creating information that wasn’t there.

His biggest takeaway wasn’t that AI is ready to do everything for us.

It’s that the people who learn how to work with it now have an enormous opportunity.

We’re Still in the Early Days

Patrick compared where we are with AI today to the early days of smartphones: enormously capable, clearly transformative, but still a long way from where the technology will eventually go.

This means there’s no reason to wait until you’re an “AI expert” to begin.

In fact, one of Patrick’s simplest pieces of advice during the conversation was:

  1. Find a problem.
  2. Try to solve it.

Pick something:

  • Repetitive
  • Frustrating
  • That takes too much time
  • Or simply something you’re interested in

Then start experimenting.

You don’t have to build an AI platform. You just have to build your own understanding of what AI does well, what it doesn’t, and how to communicate with it more effectively.

Five Prompts to Help You Get Started

You can copy, paste and customize any of these in the AI platform you already use.

  1. Find something worth solving.

    Not sure where AI could actually help you?

    Why try it: The best place to start with AI isn’t necessarily a flashy use case. Start with a real problem.

    Here are some of the things I do in a typical week: [DESCRIBE YOUR WORK OR PASTE A LIST]. Ask me questions one at a time until you understand my workflow. Then identify five tasks where AI could save me meaningful time or improve the quality of my work. Start with the opportunities that would be easiest for me to test.

  2. Make AI ask questions before it answers.

    Better input usually creates better output.

    Why try it: Treat AI less like a search box and more like a collaborator that needs a good brief.

    I need your help with [DESCRIBE THE TASK]. Before you begin, ask me the questions you need answered to do this well. Ask one question at a time and don’t begin the work until you have enough context to produce a strong result.

  3. Ask AI to challenge its own work.

    AI can sound very confident, even when it shouldn’t.

    Why try it: A polished answer isn’t necessarily an accurate answer. Patrick’s experience building Audience ID repeatedly demonstrated the importance of verification and human oversight.

    Review the answer you just gave me as a skeptical expert. Identify any assumptions you made, information you could not verify, and three areas most likely to contain an error or need additional human review. Do not change the answer yet. Show me the potential problems first.

  4. Give it boundaries.

    When accuracy matters, tell AI what it is and isn’t allowed to use.

    Why try it: Sometimes AI tries a little too hard to be helpful. Clear boundaries can make the output much more useful.

    Complete this task using only the information and sources I provide. Do not add facts, statistics or assumptions that aren’t supported by those sources. If something needed to answer the question isn’t available, identify the gap rather than filling it in.

  5. Diagnose before you iterate.

    When the result isn’t right, don’t immediately start over.

    Why try it: Patrick found that one of the keys to working effectively with AI was understanding the root of a problem instead of repeatedly making superficial fixes.

    This is the result I received: [PASTE OUTPUT]. Here’s what isn’t working: [EXPLAIN]. Before revising it, diagnose the likely root cause. Tell me what you think went wrong, what you recommend changing, and why. Then make the revision.

The Bigger Opportunity: Know Your Audience Better

When building AudienceID, Patrick set out to solve a real marketing problem to help our agency better analyze, build, and show off our capabilities.

Traditionally, developing a detailed audience strategy can require multiple data sources, manual analysis and weeks of work. Audience ID brings consumer data, audience planning and AI together into a single environment designed to help marketers move from a question about their customers to an actionable audience strategy.

During Tulsa Tech Week, Patrick demonstrated how the platform can help marketers:

  • Identify and prioritize target audiences.
  • Explore the behaviors and characteristics that distinguish them.
  • Develop richer audience personas from thousands of data points.
  • Understand how audience concentrations change by geography and location.
  • Explore business scenarios and growth opportunities.
  • Inform which messages and creative approaches should be delivered to different audiences.

In one demonstration, the platform analyzed target audiences around hundreds of individual store locations, making it possible to see how the best audience and messaging opportunity could differ from location to location rather than applying one national approach everywhere.

And that’s where AI gets especially interesting to us.

At AcrobatAnt, our Humantity™ methodology starts with Connection: understanding the values, beliefs and motivations of the people a brand hopes to reach and then aligning the brand accordingly.

Because finding someone isn’t the same as connecting with them.

Technology can help us understand an audience at a scale and speed that were not previously possible. But data alone doesn’t create relevance. Creative alone doesn’t guarantee connection. And AI alone doesn’t replace human judgment.

When we bring the intelligence, strategy and creative together, the opportunity gets much bigger.

Want to See What Patrick Shared?

Whether you joined us at Tulsa Tech Week or couldn’t make it, we’d be happy to send you Patrick’s presentation.

Request the presentation

We’ll send the presentation to the email address you provide.

Want to Go a Step Further?

Maybe the question isn’t simply, <em>“How should we use AI?”</em>

Maybe it’s:

  • Who are our best customers?
  • What actually motivates them?
  • Are we reaching the right people?
  • Does our creative align with what matters to them?
  • Could better audience intelligence improve our marketing performance?

If you’d like to explore those questions for your organization, tell us. We’ll follow up to learn a little more about what you’re trying to solve and whether our audience intelligence capabilities could help.

Yes, I’d like to learn more about audience intelligence and how it could apply to my organization.

Don’t Just Use More Technology. Use It to Matter More.

AI is changing what’s possible in marketing. But more data, more automation and more technology don’t automatically create better marketing.

The opportunity is using these tools to understand people better, make smarter decisions, and create work that connects.

  1. Connection
  2. Community
  3. Conversion

That’s the thinking behind AcrobatAnt’s Humantity™ methodology, combining human insight, audience intelligence, strategy, creative and technology to help brands matter to the people who matter to them.