The Mental Backlog

There’s a version of this blog post where I explain the market size, the LLM tailwinds, the go-to-market thesis, and why agentic AI is finally ready for prime time.

But that’s not why we backed Nir and Yoav.

One thing stuck when Nir and I had lengthy conversations about what he wanted to build. Nir said: “Every executive has a mental backlog. The scheduling email they’ll get to later, the travel they haven’t booked, the follow-up they keep meaning to send. The backlog doesn’t go away. It just accumulates. We’re going to clear it.”  That’s not just a product pitch. That’s a deep understanding of human psychology.

As I am quite a busy person (I think), I run the fund, I sit on more than a dozen boards, managing a portfolio across three continents. My schedule is a negotiation that happens mostly without me in the loop. Every Managing Partner I know lives some version of this.

So does every VP of Sales, every Head of Product, every CTO past a certain scale. The admin doesn’t go away when you get more senior. It compounds.

Illustration: AI

The problem nobody solved

Here’s what we think the market gets wrong: most decided this was a scheduling software problem. Calendly has 20 million users. It works fine for inbound booking. But it doesn’t reschedule when your flight gets delayed. It doesn’t notice you booked a red-eye with a 7am board meeting on the other end. It doesn’t draft the follow-up email from the conference.

Because executive admin isn’t a volume problem. It’s a judgment problem.

Scheduling is knowing which meeting to protect, who deserves 30 minutes vs. an hour, how to sequence a day without context-switching yourself into ineffectiveness. Travel is knowing your seat preference, your hotel radius, loyalty, that a certain routing will destroy you. Follow-ups are tone, timing, and knowing what’s worth chasing.

A great Executive Assistant learns all of this over years. EAs for middle management in the USA cost $80K–$180K annually, which is why typically the CEO gets one and possibly some C-levels over time. Everyone else manages the admin themselves and executives effectively lose a week a month to this.

Every tool that tried to solve this so far failed at the same point: it could handle volume, but it couldn’t handle judgment. Some tools have tried to be too wide, with the result that most are not really good at any one thing. 

What they build and why it’s different

The approach that an agentic personal assistant is one comprehensive product is fundamentally flawed.  There will be a dozen agents covering different aspects of any one job.  Catch understands this and has taken the approach of building the best solution for a specific set of use cases of EAs.

Catch is not a tool or a SaaS. It’s not a copilot that waits to be asked.

It’s an AI admin employee fulfilling a subset of functions, one that onboards in minutes, learns your patterns without being configured, and starts making judgment calls on your behalf before you’ve thought to ask.

The architecture behind this matters. Nir and Yoav built what they call a splinter-agent system: each task is handled by a scoped agent that only touches what it needs to. Quality goes up, latency goes down, risk stays controlled. The underlying models are fine-tuned specifically for executive admin behavior, not general-purpose, not off-the-shelf. SOC 2 compliance was built in from day one, because this is high-trust data. And critically, the system learns from every interaction, getting better at being your assistant over time.

The user experience, across WhatsApp, email, Slack, or phone, is designed to feel like a great human assistant, not like talking to a bot. Not another app on the phone, not a unique UI. Rather embedded in the communication apps customers use today.

Catch had already scheduled 12,000+ meetings, delegated 20,000+ tasks, and handled 200,000+ emails for early users.

The team

I’ve known Nir for years, he was an investor at Entrée where he focused on AI investments. He came to this challenge from three angles: operator, investor, and executive who personally lived the admin burden. 

Yoav Ramon was CTO at Nym Health and first employee at HiAuto. He’s been building AI-native systems since before it was fashionable, and understands not only the tech – but also how to make it work in the real world. The architecture reflects someone who has thought seriously about deploying AI in high-trust environments.

Why now and why niche wins

As I wrote above, the dominant narrative is that the future belongs to general-purpose agents. One AI that does everything. We believe otherwise.

The future is very niche. Agents that are genuinely exceptional at one specific domain, trained on the right data, fine-tuned on the right behaviors, optimized for the judgment calls that matter in that context. General is mediocre at everything. Specific is excellent at a few things. In professional contexts, excellent at one thing is what you actually need.

Catch’s bet is on specificity: to be, ultimately, the best executive admin employee that has ever existed, in every company, for every executive, from day one.

We co-led the $5M seed alongside Pitango, with Seedcamp and Factorial Capital. We are proud to partner with Nir and Yoav as Catch comes out of stealth. Catch is already growing revenues, customers, successful tasks completed and other KPIs ahead of expectations. 

The admin era of AI agents has begun. And Catch is leading it!

Read the PR announcement here

Catch is available now at catchagent.ai