AI Marketing That Brings in Customers: Lessons From Startup Folsom’s Panel Discussion

When Jason Wang started building an AI system to take restaurant orders over the phone, it sometimes gave customers two cheeseburgers when they had ordered one. It also added things they hadn’t requested. The restaurant owner had a reasonable question: who was paying for the extra food?

Jason said his team spent substantial time adding rules and limits so the system could handle the job reliably. Later, when they tried handing their own marketing over to AI, that didn’t automatically work either.

Those experiences helped set the tone for Startup Folsom’s November 10 AI for Marketing Panel Discussion at Frequency Coworking in Rancho Cordova. The conversation brought together Jesse Delgado of Storyline Media Group, Justin Mendez, who discussed his work on Arthur AI following his time at GreyDot Marketing, and Jason Wang of Cater AI and Votrix AI.

We talked about generating content, finding customers, and where automation needs more direction. The panelists had different approaches, but their examples kept returning to a practical question: what is this doing for the business?

Here are seven lessons from the discussion.

1. Know the process before you automate it

Jason’s restaurant example made the stakes easy to understand. An AI that sounds convincing on the phone still needs to get the order right.

He recommended establishing standard operating procedures before putting AI into customer service. That means documenting what it should do, what it should avoid, and the requirements for a successful interaction.

The same applies to marketing. A business owner needs to explain who the customer is, what the business offers, and how an inquiry should move toward a sale. Without that direction, automation has gaps to fill.

My suggestion is to write down the existing workflow before shopping for a tool. If a new employee couldn’t follow your instructions, an AI system will probably need more information too.

2. Measure the inquiries and sales your marketing produces

Justin challenged a familiar marketing report: website traffic went up, so the campaign must be working.

He described agencies and business owners focusing on informational blog content because it can generate impressions and visits. His follow-up question was whether those visitors actually called the business or became customers.

During the discussion, I admitted that I have also created blogs to try to generate traffic. It’s easy to understand the appeal. You publish something, watch the numbers increase, and feel like you’re making progress.

Justin encouraged businesses to pay attention to search terms that suggest someone is looking to buy. For example, someone searching for “emergency plumber Rancho Cordova” is in a different situation from someone researching how household plumbing works.

Educational content can still serve a purpose. The useful question is how it helps the customer and supports the business. Track qualified inquiries, booked appointments, and sales alongside traffic so you can assess that connection.

3. Give AI access to what makes your business different

Justin described building his marketing experience and documented processes into Arthur AI. Part of his approach involves interviewing the business owner to learn what they do, whom they want to serve, and what they know from experience.

His example was a plumber who handles water heater removals. That owner has practical knowledge about the work that a broad request for an SEO article won’t capture on its own.

This is something business owners can apply regardless of which tool they use. Before asking AI to write, give it material worth writing about:

  • Questions customers repeatedly ask

  • A specific job or project and what happened

  • Common mistakes you help customers avoid

  • How your process works and why you use it

Then check whether the draft actually answers the customer’s question. Asking for “good content” leaves too much undefined. Explain what the reader should understand or be able to do after reading it.

4. Some useful AI work happens behind the scenes

Jesse talked about using AI for list building, contact verification, and separating contacts into relevant customer groups. He also emphasized tracking which marketing activities lead people to take action.

Those tasks may be less exciting than generating a video, but they address real sources of wasted effort. An outdated contact list or an offer sent to the wrong audience can undermine a campaign before the creative gets a fair chance.

He also discussed simple automation opportunities for service businesses, including following up by text after a missed call.

I think this is a useful place for many small businesses to look first. Review where interested customers fall out of your process. A faster response to an existing inquiry may be more useful than another batch of social posts.

5. Match the content format to the audience and purpose

The panel didn’t fully agree on the future of AI-generated video, which made the conversation more useful.

Jesse argued that real people and documentary storytelling could become more valuable as synthetic content becomes easier to produce. He also emphasized understanding how the intended audience feels about AI.

Jason offered a different experience. He said some of his team’s AI-generated video tests attracted more traffic than their attempts to film themselves. He also discussed experiments with short dramatic stories that kept viewers interested despite imperfect visuals.

Those examples suggest a practical approach: test a format against a defined goal. Views can tell you something about attention. Inquiries and sales tell you more about the commercial result.

For a business whose owner is central to the customer relationship, a straightforward video of that person explaining their work may be worth testing. An animation or explainer might help demonstrate a process that’s difficult to show on camera. Choose based on what helps your customer understand the offer.

6. Give each piece of content a job

During the audience Q&A, a question about explainer videos led to a discussion about where content belongs in the sales process.

Justin suggested that an explainer could help educate someone before a sales call or reengage a person who hadn’t yet converted. Jesse discussed how stories can hold attention and introduce an offer after giving the viewer something interesting or useful.

For a service business, that might mean a short video about a common customer problem, a page explaining the service, and an explainer sent before a consultation. Each piece answers a different question.

Before creating the next post or video, decide who it is for and what should happen next. That decision gives AI a clearer assignment and gives you a better way to judge the result.

7. Fix one constraint and measure the change

When the conversation turned to tools, Jesse encouraged owners to first understand the constraints in their businesses. He recommended addressing one at a time and collecting measurable evidence that the change works.

Justin likewise pointed to the importance of knowing what is already happening in the business. He mentioned tools such as Google Business Profile, Google Search Console, and Google Analytics as starting points for understanding visibility and performance.

Jason added a memorable reminder about the underlying offer. He described the difficulty of marketing a restaurant whose fried chicken was consistently dry. Better promotion would still leave customers eating the same chicken.

That is worth keeping in mind before spending money on an elaborate AI setup. Identify whether the problem is getting discovered, responding to inquiries, converting prospects, or delivering an experience customers want to repeat. Each calls for different work.

Put one idea into practice

Choose one measurable problem this week. It could be missed calls, slow responses, or website visitors who rarely request an estimate. Record what is happening now, define the improvement you want, and test one change.

Give the AI workflow clear instructions and review its output. Compare the result with your starting point, including the time and effort required to run it. Use that evidence to decide what to improve next.

Thank you to Jesse Delgado, Justin Mendez, and Jason Wang for sharing their experiences, and to the City of Rancho Cordova and Frequency Coworking for supporting the Rancho Cordova AI for Business Workshop Series.

Find upcoming workshops and opportunities to meet local founders at StartupFolsom.org.


About Jesse Delgado

Jesse Delgado is the founder of Storyline Media Group, a Sacramento-based video marketing agency, and a leader in Sacramento’s growing AI community. He helps businesses and founders combine storytelling, content, and AI to improve their marketing and turn ideas into measurable results.

About Justin Mendez

Founder, JustinMendez.ai · AI for builders and business leaders · Building systems so more people can build the world they want to live in.

About Jason Wang

Jason Wang is the President of Cater AI and Muncho.

Next
Next

AI for Marketing: Why Your Human Edge Matters More Than Ever