Stripe blocked a customer payment. Here's what it taught me about AI that nobody talks about.

Yesterday, a customer tried to pay me, and Stripe blocked the payment.

Today, it happened again.

One of these customers had already paid me many times with the same card. Almost a thousand dollars total. Real customer. Real business. Real relationship. No fraud. No problem before.

But suddenly, Stripe said the payment was too risky.

I reached out to support. They looked into it. Their answer was basically: "Have the customer try another card."

That sounds simple when you are only looking at a payment on a screen.

But it feels different when you are the founder. When you know the customer. When you know their business. When you know they have paid before. When you know they are not trying to scam you.

And you know that asking them to try again creates friction that should not exist.

That is what bothered me. The system made a decision without the full story.

I am not upset that fraud protection exists. Fraud protection matters. Security matters. I use AI every single day. Conecta would not exist without it. I believe every business owner needs to start learning how to use it.

But this Stripe situation made me stop and think.

When does security become too much security? When does automation become so focused on protection that it starts hurting real businesses? When does a system become so smart that it stops seeing the human being behind the payment?

My customer was not fraud. My customer was a real person. A real business owner. Someone who had already paid before.

But the system did not see the relationship. It did not see the history. It only saw risk.

I found out this was connected to Stripe's newer AI-driven Radar risk settings. More automation. More fraud prevention. More machine learning.

I understand why that exists. But it made me think about my own customers.

Most of the people I serve are small business owners. Many are Latino. Many are immigrants. Many are Black and brown business owners. Many do not fit perfectly into the financial profile that big tech companies and financial systems like to see.

That does not make them bad customers. If anything, those are the exact customers who need better tools, better platforms, and a stronger voice.

A generic risk model does not know the business owner who had a hard year last year but grew this year. It does not know the Latina business owner who just finished tax season and is using her card more than usual. It does not know the tax pro who is finally starting to get ahead.

It only sees patterns. And patterns without context can hurt people.

At the same time this was happening, I have been working on Sofia.

Sofia is the AI agent we are building inside Conecta. She is getting close.

Sofia is not a chatbot sitting on top of a website. She is deeply connected to everything Conecta does, appointments, payments, documents, e-signatures, messages, follow-ups, workflows, client communication. Before I wrote the first line of code for Conecta, I already knew the platform would need to become intelligent. It could not just be a CRM.

But after what happened with Stripe, I started looking at Sofia differently.

I am not just watching AI affect small businesses from the outside. I am building AI that will affect small businesses from the inside.

That comes with responsibility. And it forced me to ask myself a question I have been sitting with:

Am I putting my own biases into Sofia?

I grew up watching how Latino small business owners actually operate. I know the culture, the seasonality, the financial reality, the way trust gets built in this community. I have spent a decade in the tax industry. I have seen what happens when a tool does not understand the person using it.

So yes, my biases are in Sofia. The question is whether they are the right ones.

When small business owners turn this on and say, "Bert, just set it up for me," how will this affect their business? Their employees? Their customers?

Because here is the truth: AI will take jobs. A receptionist who answered calls, scheduled appointments, sent reminders, and handled basic questions may not be needed in the same way anymore. Maybe that person moves into a more important role. Maybe the owner invests in them differently. Or maybe the owner decides they do not need as many people.

That is the reality. And I have to be honest about it because Sofia is not just an idea anymore.

AI is not going away. Small businesses are going to use it because they care about time, money, missed calls, and staying alive. That part is simple.

But the Stripe situation reminded me that the future cannot just be automation for the sake of automation.

When systems make decisions without human context, real people get hurt. A payment gets blocked. A customer gets embarrassed. A business loses revenue. A founder loses trust in a platform. And everyone is left saying, "The system decided."

That is not good enough.

The more AI grows, the more valuable human judgment becomes. The more automation we add, the more important empathy becomes. The more systems we build, the more we need relationships.

I used to think those were soft ideas. I do not think that anymore.

The businesses that win long term will not be the ones that remove humans completely. They will be the ones that combine AI efficiency with human context.

That is what I want Sofia to become, not just another AI feature, not just an automation layer, but a tool that helps small businesses grow without losing the relationships that got them there.

Sofia is coming soon. But more importantly, she is being built with this responsibility in mind.

Because small businesses do not just need more automation.

They need automation that understands context. They need tools that help them grow without losing the relationships that got them there.

That is the future I want Conecta to help build.