Why I Founded InfiniteWatch: Building the Trust Layer for the Agentic Internet

After spending thousands of hours manually monitoring customer interactions at CoverWallet, I knew there had to be a better way. Today I'm announcing InfiniteWatch.

I've spent thousands of hours watching session replays and listening to customer phone calls.

Not because I wanted to. Because I had to.

At CoverWallet, the insurance tech company that I made grow to over $1 billion in premium revenue before being acquired by Aon, understanding our customers was existential. Every friction point in the application flow meant lost revenue. Every confusing interaction on a call meant a customer walking away.

So we did what every company does: we watched. We listened.

My co-founders and I would spend too many hours reviewing session recordings, trying to spot patterns in how users struggled with our product. We'd listen to call recordings, noting where customers got confused or frustrated. We'd compile spreadsheets of issues, manually categorizing problems, estimating impact.

It simply didn't scale. And we knew we were only seeing a fraction of what was actually happening.

The Breaking Point

Here's the thing about manual monitoring: it's not just inefficient, it's fundamentally impossible at meaningful scale.

You can't manually review 10,000 daily interactions and catch every bug, compliance issue, or revenue leak. The insights come too late. You end up sampling randomly, hoping you catch the important stuff, knowing you're missing most of it.

Even worse, by the time you identify a problem, hundreds or thousands of customers have already experienced it. You're always playing catch-up, always reactive, never truly in control.

I remember we used to think: There has to be a better way.

The Realization

That better way is AI.

Not AI that flags rage clicks or creates automated summaries as an afterthought.

I mean AI as the fundamental intelligence layer that watches every interaction, understands context, identifies patterns, and surfaces insights, all automatically, at scale, in real-time.

The technology finally exists to do what we used to do manually, but better: semantic understanding of user behavior, automatic detection of friction points across modalities, intelligent prioritization of issues by business impact.

That's why my co-founders Inaki Berenguer, Alberto González and I started InfiniteWatch.

But Here's the Real Opportunity

As we started building, we realized something bigger was happening.

We're entering the era of the Agentic Internet.

In the next 2-3 years, trillions of daily interactions will unfold between AI agents and users, and between agents themselves. AI voice agents handling customer service calls. AI browsing agents navigating websites to book travel or make purchases. AI co-pilots guiding users through complex workflows.

Companies are already deploying these systems. They're promising efficiency gains, cost reduction, 24/7 availability.

But here's what no one is talking about: Companies won't fully trust these AI agents without complete visibility into what they're doing.

Think about it. Would you hand over your customer service operation to an AI agent if you couldn't monitor QA, compliance, user experience, revenue leakage, customer satisfaction, reputation, … Of course not.

The Agentic Internet needs an observability layer. It needs trust infrastructure.

That's what we're building at InfiniteWatch.

What We've Built

I'm excited to announce that InfiniteWatch is coming out of stealth with $4 million in pre-seed funding led by Base10 Partners, with participation from Sequoia and A16Z scouts, Kibo Ventures, Kfund, LifeX, and a list unicorn founders.

Our platform has three core products as AI agents today, with more coming in 2026:

AI Web Insights: Automatically analyzes web sessions and surfaces the exact UX issues killing conversions. Within 24 hours of deployment, one of our customers (TheGuarantors) had 10x more visibility into critical conversion blockers and increased conversion rates within the first week.

AI Voice Listener and Simulator: Deploys synthetic personas to test voice AI systems at scale, uncovering edge cases before customers encounter them. In listening mode, it continuously analyzes calls for insights on performance, compliance, and customer satisfaction.

Coming in 2026: AI Web Testing and AI Co-Browsing capabilities, all sharing intelligence across a unified platform.

But here's what makes InfiniteWatch different: we don't just monitor. We understand.

Our AI watches sessions and listens to calls the way a human analyst would, but at scale and in real-time. It connects what happens in web sessions with what's said on phone calls, creating a complete picture of the customer journey. It tells you not just what happened, but why it matters and how to fix it.

Why Now

I couldn't have built this company five years ago. The AI technology simply wasn't ready.

But today, with advances in large language models and semantic understanding, we can finally automate what used to require human judgment. We can process multimodal data (video, audio, text) at scale. We can identify patterns that no human could spot across thousands of simultaneous interactions.

And the timing matters because companies are deploying AI agents right now. They need observability right now. They need trust infrastructure right now.

We're not building for a hypothetical future. We're building for the reality that's already unfolding.