🎧 Práctica de pronunciación

Sure. My background has two halves that ended up connecting.

The first ten years I spent producing visual content professionally. I started as a photographer, worked on international cruise ships across Europe and Asia, and then founded and ran my own commercial photography studio in Chile for four years. So I directed shoots, I edited, and I ran the business side of it too.

The last three years have been different. I joined an e-commerce group in Medellín to do marketing, and I ended up building their whole automation and AI infrastructure. Today they run more than ten brands on systems I built. AI agents that handle sales conversations on WhatsApp. A reporting platform the leadership team opens every morning. Content pipelines that generate assets with AI. And along the way I led a team of eleven people across design, video, community and paid media.

So I've been on both sides of this problem. I know what makes a piece of content actually work, because I made content for ten years. And I know how to build the system that produces it, because that's what I do now.

The way I work hasn't really changed across projects. I map the operation first, before I touch anything. I sit with the people actually doing the work, not just the person who hired me, and I look for where time is being lost. Then I decide what deserves to be built, I build it, I ship it, and I measure it.

There's one rule I follow. Anything I find myself doing twice, I turn into a system.

And I care a lot about what happens after launch. A system nobody uses a month later is a failed system. So what I build comes with monitoring, it comes documented, and it keeps running without me watching over it.

Here's an example. While I was building the reporting platform, I noticed the numbers from the sales platform and the numbers from fulfillment never quite matched. Nobody asked me to look into it. There was no ticket, no complaint. It was only visible if you crossed both sources.

So I built an automatic reconciliation between the two systems, comparing order by order instead of looking at totals.

It turned out about twenty five percent of orders were stalling in a gap between the two systems. It wasn't a software bug. It was a process gap nobody was watching, because each team only saw their own side, and their own numbers looked fine.

What I took from that is that the expensive problems usually aren't inside a process. They're in the seams between two processes, where nobody owns the outcome.

The hardest part of the e-commerce work wasn't technical, it was the simultaneity. Every brand had its own catalog, its own tone, its own operation, and building one system per brand wasn't going to scale.

The design decision that solved it was making the agents template-replicable. I separated the shared logic from what's specific to each brand. So the routing, the memory, the escalation rules, all of that is common. Only the catalog and the voice change.

The result is that launching a new brand went from weeks to days. And there are more than ten of those agents running in production right now.

One thing I built that's close to what you're describing here. For the e-commerce brands I set up a content pipeline where the ads that performed best fed back into the prompts that generated the next batch. So the system got better every cycle instead of starting from zero each time.

That feedback loop is the part most people skip. It's easy to build something that produces a lot. It's harder to build something that learns from what actually worked. But without it, you just have a factory that's blind.

I led a team of eleven across design, video, community management and paid media. Part of that was workload allocation and quality review. But the bigger part was removing the bottlenecks between them.

That's usually where the time actually gets lost. Not in the work itself, in the handoffs. Design waiting on copy, paid media waiting on assets. So a lot of my job was making those handoffs visible, and then removing them.

What caught my attention is that this isn't a content production role, it's a systems role. The output is content, but the actual deliverable is the engine that produces it.

That's close to how I already work. And honestly, it's the first role I've seen that needs both halves of my background at the same time. You need someone who can look at an hour of recording and know which sixty seconds are worth clipping. And you need someone who can then build the system that finds those sixty seconds without a human watching the whole hour.

I've spent ten years on the first part, and three years on the second.

Sorry, could you repeat that?

Would you mind rephrasing that?

I want to make sure I understood. Are you asking about the technical side, or the process side?

That's a good question, let me think for a second.

Let me give you a concrete example.

I haven't worked with that specific tool, but here's what I have done that's close.

I don't want to guess on that one. What I can tell you is this.

There are two parts to that.

The way I'd approach it is this.

And the result was…

That makes sense.

That's exactly the problem I ran into.

Thank you, this was really helpful.

I'm genuinely excited about this one.

Looking forward to the next step.

workflow · pipeline · to ship · in production · to scale · bottleneck · stakeholder · to own · ownership · deliverable · roadmap · throughput · turnaround time · leverage · to surface

Y la frase que más vas a decir:
I've spent the last three years building AI and automation systems in production.

What does the content intake look like today? Are recordings already being captured systematically, or would building that be the first thing?

You mention five hundred pieces a month by day ninety. What's the current baseline you're working from?

How does the engineering team work today? Would I be writing specs for them, or building the first versions myself in tools like n8n?

Since this is a founding role, how much freedom would I have to define the workflow, versus following something that already exists?

What does success look like for you personally in the first ninety days?