AI has become the loudest voice in product conversations. Every week brings a new hyperbolic claim that AI has “replaced” product managers or made strategy “obsolete.” But beneath the noise, the real truth is emerging. 

The teams that win in the AI era are not the ones shipping the most agents. They are the ones with the deepest understanding of what they are building, who they are building for, and why it matters. You still need product managers to get this right.

In the latest episode of Untrapping Product Teams, Productboard founder and CEO Hubert Palan explains that while AI accelerates product work, it doesn’t replace judgment. As building becomes easier, correctly deciding what to build separates the winning teams from everyone else.

That belief has guided Productboard since the beginning. It’s only been more proven with AI.

AI Makes Product Specifications Even More Valuable

Much of the anxiety around AI assumes that automation eliminates the need for product thinking. Hubert argues the opposite. As AI systems take on more execution work, the quality of direction matters more than ever.

AI can generate code and summarize feedback. What it cannot do is decide what problem is actually worth solving. That responsibility still sits with humans. When teams lack clarity, AI simply accelerates the wrong outcomes faster.

As Hubert put it:

“Specs are becoming the new code. If you assume AI agents are going to ship the code, the real question is, ‘What do you tell the agent to ship?’ What’s the specification? We’re not in a world where you can just say, ‘AI, figure out the problem and the solution.’”

Specifications, intent, and shared understanding are becoming more critical than code itself. Product managers and engineers are increasingly working in a single, compressed workflow. But without deep understanding upfront, speed won’t translate into impact.

Focus Is the Must-Have Product Manager Skill

AI makes it cheaper to build, but it does not make choosing easier. If anything, abundance increases the cost of poor prioritization.

Hubert reflected on how Productboard has navigated this challenge over time. Each inflection point required focus and conviction:

• Deciding which customers to serve.
• Resetting roadmap detours driven by individual large accounts.
• Investing early in AI—before every customer was ready.

The pattern is consistent across enduring companies; winning organizations do not try to serve everyone. They’re opinionated about who they build for and methodical about how they allocate resources. AI now makes this unavoidable.

Why Most AI Startups Fail

Most AI startups fail because they do not know their market deeply enough.

The rapid rise of these AI startups has created the illusion that technology itself is the differentiator. In reality, many of the most successful AI products come from teams with deep domain expertise, not just technical skill.

Engineers building tools for other engineers succeed because they understand the workflow intimately. That understanding shapes product decisions long before any model is trained or deployed. By contrast, teams that attempt to apply AI to unfamiliar domains struggle to even evaluate whether AI outputs are correct.

Without market mastery, feedback loops slow down. Validation becomes expensive and, suddenly, all that speed disappears. This is why context is such an advantage. Teams who understand their market deeply can move faster precisely because they do not need constant external validation to know whether something makes sense.

Generic AI Tools Cannot Provide Context

General-purpose AI tools are powerful, but they operate in isolation. They lack shared and persistent organizational context. 

Product decisions rarely depend on a single input. They require customer feedback, competitive insight, internal strategy, historical learnings, and an understanding of what has already been tried. When this context is fragmented across tools and teams, AI outputs remain shallow.

This is where specialized systems outperform generic assistants. AI becomes meaningfully useful only when it operates inside the product environment where context already exists and decisions are made collaboratively.

This challenge is exactly where Productboard’s AI work is focused.

Productboard Spark is designed to operate within real product context, not outside of it. It brings customer insights, competitive signals, internal strategy, and historical decisions together in one shared system. Rather than generating answers in isolation, Spark supports the actual product workflow, from discovery and synthesis to prioritization and specification. Purpose-built for product managers, this AI agent turns real context into PRDs, competitive analysis, and even launch plans.

Spark scales product thinking. By keeping context persistent and decision-ready, Spark helps teams move faster without sacrificing clarity. 

The Future of Product Work: AI-Centered, Human-Augmented

Despite the pace of change, one principal remains constant: humans are still accountable for product outcomes.

AI can accelerate discovery and surface patterns across massive datasets, reducing the time it takes to move from insight to specification. But alignment, judgment, and ownership do not disappear. Product teams still need to internalize what they learn, debate tradeoffs, and build shared understanding across the organization.

In an AI-saturated world, context beats technology every time. The teams that invest in understanding their customers, markets, and internal knowledge systems will outpace those chasing speed alone.

AI will shape how products are built. But context will decide which ones win.

Drive real product outcomes with Spark.

• Join the waitlist for our context-driven agent.
• Watch the full episode.