Tackling the Challenges of Product Development in the AI Era

As AI-enabled tools speed up software development, startup product leaders are increasingly focusing on the fundamentals.

In our view the formula for building a compelling product that solves high-value problems has been clear for years: deep alignment around a clear business strategy; a ruthlessly prioritized roadmap to deliver on that strategy; and fast decision-making, even in the face of ambiguity. The shift to AI-enabled software development is making this playbook more important than ever. For the latest installment in our series on the trends shaping the modern startup ecosystem, EQT’s Early Stage Operating Team interviewed executives at companies from Harvey, a legal AI business, to Fin, which uses AI to build customer support tools, on how they’re adjusting to ever-faster production timelines.

Clarity is king

AI can mean you now have the potential to build almost any software, almost instantly. That should feel like freedom. Instead, it forces a reckoning.

For decades, we have seen product leaders preaching discipline and focus. Pick your bets. Say no. Concentrate resources. But the reality is that traditional software companies moved slowly enough that mediocre prioritization could hide under layers of process. You could ship 10 features a quarter and easily confuse quantity for quality. You could make up for lack of focus with more meetings, more review cycles, more time for consensus-building.

“Shipping a bunch of features is easy,” says Anique Drumright, chief product officer at Harvey. “Taking the time and being disciplined about how to use the capacity you have on a small handful of big bets that meaningfully move the needle for customers and for the business is critical, but harder than ever."

Previously, the constraint was execution capacity; now it’s strategic clarity.

This changes how product roadmaps work, according to Des Traynor, chief strategy officer at Fin, the company formerly known as Intercom. Instead of 12-month plans, you're working in six-month windows – or even quarterly delivery plans with “directional guidance” beyond that. "You need maximum flexibility, so roadmaps can't really go beyond six months and even then it has to be OK within the culture to rip it up and make new plans," he says.

But that flexibility only works if you know why you're building what you're building. Every sprint has to be oriented toward a clear objective, or you're just shipping noise.

Make haste, slowly

Speed doesn't reduce the need for clear company goals. In fact, it multiplies it. Moving quickly means decision-making is pushed down an organization, creating an urgent need for shared context. Without it, distributed decisions can become incoherent ones.

Drumright introduced us to a concept that captures this perfectly: "context leaking". This is the practice of continuous sharing of information, up and down a management chain, across functions and departments. It requires time and intention, but it's critical to how the Harvey team trusts that everyone can make decisions without bureaucratic, time-consuming processes.

Everyone needs to know why decisions are being made, what the customer context is, what the market looks like, and how it all drives the key business priorities. That information has to flow constantly and in all directions, she says.

"What you need is a shared north star,” says Patrik Torstensson, head of engineering at Lovable, an AI app developer. “We talk a lot about one plus one equals three. That doesn't happen if people are siloed." This is the inversion of the old model. Hierarchy used to control decision-making. Bureaucracy was the cost of coordination. Now startups are replacing hierarchy with shared context and ritual. More communication, not less. Speed comes from trusting people to make decisions – but that trust has to be earned through relentless, synchronous information sharing.

More isn’t always better

While LinkedIn is littered with posts about “how I built a CRM in a weekend with 800 agents and a small fortune in tokens”, when features can be built quickly winning isn’t about how many features you can ship but how well those features come together to solve real business problems for users.

When a designer can publish a user interface change, when a product manager can fix a bug, when engineers can make product decisions, who's thinking about how these changes all fit together? In traditional organizations, it was architects and design systems committees. As speed removes these guardrails, everyone has to understand the whole system.

“Pace without clarity can easily create chaos,” says Lovable’s Torstensson. “This is something we take seriously."

This is the irony of AI-native product development: the craft of building a coherent product seems to matter more now, not less. You need stronger mental models of your system. You need better design patterns. You need clearer principles. Because you can't enforce them through process anymore, you have to distribute them through shared understanding.

Working with enterprise

Before we wrap up, a quick note for those selling to enterprise clients. As you move faster, your customers might not always keep pace.

When you can ship in days, a gap can open up between you launching new features and your clients adopting them. This means that it could become increasingly important to maintain strong customer relationships. That trust isn't just uptime, security and data integrity, it's deeply understanding their business, speaking their language, and respecting their ability to adjust.

Harvey’s Drumright explains: “The prevailing discourse about product development today assumes a business-to-consumer model: ship more, faster to win. This just isn’t true in the enterprise space. They’ve never been able to absorb that pace of change.”

Removing the hiding places

The practices that AI-native companies are doubling down on – focused strategy, distributed decision-making, coherent solutions – aren't new ideas. We just had enough process and bureaucracy to hide when we weren't doing them.

Speed exposes those hiding places. The companies adapting to this environment aren't throwing out the long-held best practices, they're practicing the fundamentals with the rigor that increased speed demands.