It’s the classic Silicon Valley test of conviction: you have an idea, you pitch it to operators and VCs, and they tell you it’s completely unnecessary.
For Adar Arnon and his co-founder at SafeBase (recently acquired by Drata), that “stupid idea” was the Trust Center—a product that solved the friction of enterprise security reviews. While everyone else was building automated SOC 2 compliance tools, they noticed a critical gap. Fixing that gap ultimately led to a massive exit.
In a recent conversation, Adar broke down what it actually means to operate in Founder Mode these days, how AI is changing engineering leadership, and why you your technical culture matters more than ever before.
Here are the biggest takeaways for technical founders and engineering leaders.
1. Find Greenfield in a Bloodbath
When SafeBase was coming out of Y Combinator, the automated compliance market (companies like Vanta and Drata) was already exploding. The conventional wisdom for a cybersecurity founder would be to build a better, faster SOC 2 tool.
Adar’s team realized the party had already started, and they didn’t want to fight in a bloodbath for SMB market share.
Instead of asking, *“How do we compete?”* they went to users and asked, *“What does your current SOC 2 tool NOT solve?”*
The answer was enterprise deal friction. When companies scale, security knowledge becomes siloed between sales, security, and engineering. Closing a big enterprise deal meant navigating an endless maze of security questionnaires. SafeBase built Trust Centers to solve this specific, hair-on-fire problem for mid-market and enterprise companies, completely bypassing the crowded SMB dogfight.
The Takeaway: If everyone agrees your idea is the top problem to solve today, you’re already late to the party. Find the unsexy, high-friction problem that the current “winners” are ignoring.
2. The Holy Grail: Enterprise Viral Loops
Product-Led Growth (PLG) is usually reserved for consumer apps or bottom-up developer tools. It rarely happens in enterprise B2B sales.
But SafeBase accidentally stumbled into a true enterprise viral loop. Because Trust Centers are externally facing, SafeBase’s early customers were sending their Trust Centers to their enterprise prospects.
The security buyers at those massive companies would see the Trust Center, fall in love with the seamless experience, and immediately ask, ”Can you connect us with whoever built this? We want it for our company.”
This dynamic pulled SafeBase upmarket incredibly fast, landing them 500 to 1,000 employee companies organically.
3. Extreme Ownership & The “AI Code Reviewer”
As startups scale, technical founders are inevitably told to step back. The traditional playbook says you hire a VP of Engineering, stop looking at pull requests, and focus entirely on “strategy.”
Adar completely rejects this. As a former elite IC, he reviewed code until his last day at the company.
Why? Because engineering culture isn’t built in a Notion doc. It’s built in the trenches.
”Trust is like playing tug-of-war with a rope... it’s built by interactions with people. You cannot let people run loose now, because in two years you’re gonna run into a wall of tech debt.”
Adar was notoriously strict on code reviews to ensure consistency and prevent engineers from over-abstracting their code. To scale his uncompromising standards as the team grew to 25 engineers, he did something brilliant: He built an AI bot that mimicked his exact code review style.
By training an LLM on his specific feedback patterns, he enforced the company’s engineering culture automatically, ensuring high standards persisted without him having to manually review every single PR.
4. AI is Shifting the Engineering Bottleneck
With the rise of AI coding assistants (like Copilot and Cursor), developers are outputting 2x more code than before. But that doesn’t mean you ship 2x faster.
According to Adar, the bottleneck hasn’t disappeared; it has just moved downstream.
“The new constraints are CI/CD pipelines and code reviews.”
AI is also changing how engineering managers operate. Instead of interrupting engineers for status updates or standups, managers can now use LLMs to summarize all the PRs merged in the last week, analyze repo health, and identify architectural drift—all without breaking their team’s flow state.
5. Ignore the “Series B Gaslighting”
Perhaps the most resonant lesson for growing founders is the pressure of the Series B stage.
As you raise more money and hire more executives, there is immense external pressure to manage the company the “standard” way. You will be told that you are too deep in the weeds, that you need to delegate core technical decisions, and that your hands-on approach is micromanagement.
Adar calls this out for what it is: gaslighting.
If you are a subject-matter expert in your product’s architecture, stepping away prematurely is the easiest way to lose a year to a mediocre executive hire. There is a fine line between empowering great people and abandoning the core competencies that made your company successful in the first place.
If you know in your gut that something needs to be done a certain way, hold your ground. “Founder Mode” isn’t a buzzword; it’s a requirement for long-term survival.
---
*Enjoyed this breakdown? Subscribe below for more deep dives into technical leadership, building enterprise software, and navigating the noise of the AI era.
