> ## Content Index
> Fetch the complete content index at: https://www.therecursive.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Want AI That Works? Start With the Boring Stuff: Martin Hýravý on Real Startup Automation
- URL: https://www.therecursive.com/martin-hyravy-on-scalable-startup-automation-ai-agents/
- Published: 2026-09-23T11:48:56.000Z
- Updated: 2026-09-23T12:22:05.000Z
- Description: Make's Martin Hýravý cuts through the hype ahead of the Startup Revolution AI Summit, revealing why founders must ditch flashy tech for dull, reliable automations that actually bring value and drive revenue.
- Author: Ana Marija Kostanic
- Tags: Interview, Agentic AI, North Macedonia, Startup Revolution AI Summit, Martin Hýravý, Automation, Events

In the current AI hype climate, every early-stage founder is under pressure to claim they are *agentic*. Autonomous workflows handle lead gen, customer support does itself, and internal ops run on algorithmic fairy dust. But, let’s be real, beneath the pitch decks, most of these setups are held together by digital duct tape.

Ahead of the [**Startup Revolution AI Summit 2026** ](https://startuprevolution.ai/?ref=therecursive.com)in Skopje this October, we sat down with **Martin Hýravý**, AI & Startup Manager at Make. Having scaled Make’s Startup Program to support over 10,000 companies globally, and spending his weeks working face-to-face with founders at Paris’s STATION F, Hýravý has seen firsthand what happens when autonomous experiments hit real-world scaling pressure.

Before presenting his upcoming keynote at the Startup Revolution AI Summit this October in Skopje, *“Agentic Systems at Scale: The Difference Between Autonomous and Unaccountable,”* Hýravý shared his pragmatic know-how for building workflow infrastructure that doesn’t collapse under its own weight.

[Startup Revolution AI Summit 2026Skopje will host one of the region’s largest startup and emerging-tech gatherings as the Startup Revolution AI Summit returns this October.![](https://storage.ghost.io/c/b8/ae/b8aee878-7ddc-40ad-8f68-92a28c560560/content/images/icon/PNG_recrop_resize_logo_only-9e756d93-1e81-44de-a3f9-9520c6858ea6.png)The RecursiveThe Recursive![](https://storage.ghost.io/c/b8/ae/b8aee878-7ddc-40ad-8f68-92a28c560560/content/images/thumbnail/money-europe-eca31ffb-fd08-48f7-9cac-bd0519932d8d.png)](https://www.therecursive.com/startup-revolution-ai-summit-2026/)

## “Don’t start with agents”

When early-stage founders approach Hýravý asking where to deploy their first AI agent, his initial advice is: **don’t**. Instead, look at the dull, manual tasks that waste your team’s hours every single week.

“*Don’t start with agents*,” Hýravý says. "*Start with whatever repetitive, critical process is costing you the most hours right now. For most early-stage founders, that’s CRM hygiene or moving data between disconnected tools. Automate that one thing end to end before building anything resembling an ‘agentic system.’"*

He points out that startups usually pour all their energy into the core product while ignoring internal tasks like data handoffs, process mapping, and basic reporting.

> “*By the time a startup hits its first real growth barrier, those unautomated internal processes become the bottleneck, and by then, fixing them costs far more than building them right from day one would have*."

For Hýravý, the primary rule is non-negotiable: "*The one that holds up is simple: every automation or agent needs to be tied to a specific, measurable outcome before it goes live. Not ‘we have agents running our ops’ as the goal itself."* 

Starting with simple tasks might feel less exciting, “*but it gives you an immediate, measurable return, and it teaches you what good automation looks like before you build something more complex on top of it*.”

## The trap of building complexity for its own sake

Why do so many startups struggle with this? According to Hýravý, founders consistently pour their energy into customer-facing products while **treating internal scaling infrastructure as an afterthought**. By the time rapid growth arrives, those neglected internal processes turn into expensive bottlenecks.

Even worse is the temptation to build architectural complexity for its own sake.

“*I once sat with a founder who’d built an entire agentic environment to manage his automations, and he was spending more time and money maintaining that environment than the automations were saving him anywhere else*,” Hýravý recalls. "*That startup shut down a few weeks before we spoke. I won’t claim the tooling caused that, but it’s a clear signal of where his time was actually going instead of the business."*

This trap isn’t exclusive to tiny startups. Even traditional enterprises struggling with legacy tech debt need to keep things focused. His advice for larger teams follows the exact same logic: “*start with one contained process, tie it to one measurable outcome, And if the legacy system, in and of itself, is a large enough barrier, then migration should be the priority number one*.”

Whether managing five people or fifty, Hýravý emphasizes one core habit: “**build for reliability over speed*. It’s tempting to ship something fast that works today; the founders who scale well are the ones who invest early in processes that don’t break when volume goes up*.”

### Measure sales velocity, not vanity numbers

The tendency to over-engineer often spills directly into *Go-To-Market* strategy. Drawing on insights he highlighted during a recent event in Amsterdam, Hýravý notes that **startups deploying AI agents into sales pipelines** regularly focus on signals that sound impressive but miss the mark.

During those critical first few weeks of running an AI sales agent, figures like “hours saved” or “tasks completed” look neat in pitch decks, but they don’t prove the business is actually growing.

“*The signal I keep coming back to is *sales velocity, not productivity**,” Hýravý explains. "*It’s tempting to measure a sales agent by ‘time saved’ or ‘tasks completed’. Those numbers look cool on a slide but don’t tell you if the agent is actually moving deals. I’d rather see a founder tie the agent to one specific component of sales velocity than report a productivity figure disconnected from revenue."*

As the company scales, that focus stays the same: track how automation drives revenue, not just how busy the AI looks.

### Preparing for the inevitable *failure*

This gap between pilot-stage promises and real production failure forms the heart of Hýravý’s upcoming case study at the Startup Revolution AI Summit this October in Skopje, titled *“Agentic Systems at Scale: The Difference Between Autonomous and Unaccountable.”*

In Hýravý’s view, **expecting an AI system never to glitch is unrealistic**. Every agent will break at some point; what matters is how you plan for that moment.

> “*Every agentic system fails at some point. It’s rather a question of whether it’s going to be safe and cheap or late and expensive*.” 

His talk will walk through the specific failure modes that appear once systems leave the test environment, offering practical ways to keep automations reliable under pressure.

## Sustainable processes don’t burn out

Hýravý’s focus on simple, sturdy systems stems from his background. Before Make, he closed major deals at Back Market—a certified B Corp selling refurbished tech—and later earned a Master’s in Impact Entrepreneurship at NOVA SBE. That background taught him to look at technical efficiency through the lens of long-term sustainability.

*“What it changed is how I think about efficiency,”* Hýravý shares*.* 

> *"Even now, working in software and AI, sustainability is a value I try to bring into everything I build, not just sustainable products, but *sustainable processes*. That second part is underrated. Everyone optimizes for what a product does; very few optimize for whether the way it’s built and run can hold up without burning out the team or the systems behind it*."

As founders prepare for the [Startup Revolution AI Summit this October](https://startuprevolution.ai/?ref=therecursive.com), Hýravý’s message serves as a grounded sanity check: **fix your basic data, track actual revenue movement, and make sure your internal processes are built to last** before reaching for the next hyped AI agent.