The Power of "And": Lessons From Building Cross-Functional Excellence at Shepherd

Shepherd’s Rise 🚀

Over the past four years, we’ve built the hottest insurtech in commercial construction — a nine-figure book of premium, a disciplined underwriting operation, and an extremely lean team. Being the first hire gave me a front-row seat to how this all came together: a rare blend of deep domain expertise in commercial insurance and a strong, engineering-driven culture.

Today, “AI-enabled services” startups are everywhere: Shepherd, Mercor, Mechanical Orchard, and others. Firms like Emergence and 8VC have published entire theses on this category. But four years ago, pre-GPT, betting on a company that didn’t sell software - and instead used software to deliver a drastically better service - was contrarian. Most people dismissed the model entirely.

Thankfully, we were contrarian and right.

That said, I’d be lying if I pitched you a hero’s-journey fairy tale. Building a tech-native services company is brutally hard. There are dozens of structural challenges, but today I want to focus on one that almost no one talks about - cross-functional excellence.

It’s the unsexy, often invisible muscle that determines whether an AI-native services startup ever reaches escape velocity. And the teams that ignore it simply won’t.

Cross-Functional Excellence: What’s that and why is it important?💡

In a traditional SaaS startup, the organizational model is straightforward:

• Engineers build a product they often use themselves (Notion, GitHub, etc.).
• Sales focuses on distributing that product.
• The playbooks are established, the feedback loops are short, and the empathy gap is minimal.

Everyone involved is close to the customer by default because, in many cases, they are the customer - an engineer naturally uses GitHub, a PM uses Notion in their own workflow. That embedded empathy quietly drives thousands of good decisions and prevents a mountain of dysfunction. It’s one of the most underrated advantages of classic SaaS.

AI-enabled services startups don’t get that benefit. In domain-heavy industries like law, consulting, or (my favorite) insurance, the people building the technology aren’t the end users and can’t rely on intuition to know what “right” looks like. At Shepherd, this was obvious in the early days: software engineers and insurance professionals lived in different worlds. They are rarely part of the same circles, hang out in the same places, or share the same mental models. This divide is the trap most AI-enabled services startups fall into. But it’s also the opportunity. If you can align these two worlds early, engineers and domain experts working as one, you create a team that’s incredibly hard to compete with.

Cross-functional excellence is what closes that gap. It enables technical and domain teams to operate with shared context instead of silos. Without it, AI-native service companies won’t rise. With it, you build a real moat. Companies that master it create a cultural moat and a cross-functional flywheel that becomes the critical ingredient that early-stage teams rely on.

How we solved for it 🤝

The key unlock for cross-functional excellence at Shepherd was reframing how builders relate to the customer.

In an AI-enabled services company, the teams building software aren’t the end users—and they can’t rely on intuition to know what “right” looks like. For builders, the fastest way to stay close to the external customer is by staying deeply embedded with the internal teams who serve them every day: underwriters, actuaries, customer success, compliance, and insurance leadership. They’re a Slack message, Zoom call, or short walk away.

When builders stay tightly connected to those teams as the company scales, you avoid the empathy decay that quietly kills most services businesses. External and internal customers are equally important at the company level - but for builders, internal teams are the primary interface for understanding how real customer value is created.

Once we internalized that, execution became clearer. Three decisions mattered most.

1. Hire for EQ and IQ from day one

Cross-functional excellence starts with the founding team and the first dozen hires. At Shepherd, we valued EQ and IQ equally on both sides of the aisle.

Early engineers had to act like product managers. They talked directly to underwriters, sat in on deal reviews, and learned the business well enough to build intuition, not just features. If someone wasn’t comfortable working closely with customers, they weren’t a fit - no matter how technically strong they were.

The same bar applied in reverse. We looked for domain experts who were curious about software, respected engineers, and wanted to learn how technology could change their workflows. Cross-functional excellence is a two-way street. If either side looks down on the other, you’re dead.

Those early hires became culture carriers. The expectations they set - how decisions are made, how feedback flows, how disagreements are handled - compound for years.

2. Optimize for proximity, not convenience

For a tech-native services company, proximity matters more than flexibility.

Shepherd has been an in-person company, five days a week. That wasn’t about optics or control - it was about speed of trust and depth of context. Cross-functional excellence isn’t built in Zoom meetings. It’s built at the lunch table, in hallway conversations, and in unplanned moments where domain knowledge gets transferred naturally.

One example that stuck with me early on: hearing our Head of Underwriting explain how he once lost a massive deal while working at a prior Tier-1 carrier because approvals had to be faxed through multiple layers of bureaucracy. That story did more to shape how engineers thought about underwriting speed and autonomy than any PRD ever could.

Empathy doesn’t scale over calendar invites - it scales through osmosis.

3. Make it official, and pay for it

Values don’t matter unless behavior changes.

At Shepherd, cross-functional excellence is formalized as a core value: “Cross the Aisle.” It’s not a slogan. It shows up in performance reviews, compensation decisions, and public recognition. At company offsites, we give awards to people who exemplify it - not just leaders, but individual contributors who consistently bridge gaps between functions.

Once we tied cross-functional behavior to real incentives, things clicked. Engineers spent more time understanding the business. Domain experts engaged earlier in product decisions. Collaboration stopped being “nice to have” and became table stakes.

To borrow from Charlie Munger: show me the incentive, and I’ll show you the outcome.

When you know it is working ✅

You know cross-functional excellence is working when the signals line up. NPS rises and shows up directly in revenue. Engineers build with real customer intuition, not guesses. Day-to-day conversations cut across functions, and internal politics stay low because teams are aligned by default.

This flywheel takes years to form, but once it’s there, it’s hard to copy and even harder to unwind. In AI-enabled services, the moat isn’t just models or workflows—it’s shared context and trust between builders and domain experts.

Ignore it and you’ll stall. Invest in it early, and it becomes a durable advantage that will get you to that big outcome.