AI as an Ally: Generative Tools in Practice

Kunal Belamkar
Product Designer
Read Time
7 min read
Published On
May 9, 2025

In today's tech landscape, it's impossible to escape new AI tools promising to revolutionize product development. From Claude's code generation to Relume's visual prototyping and wireframing to ChatGPT content assistance, the market is flooded with AI solutions promising to reduce development time, lower costs, and minimize human involvement.

As a product designer, I've had a front-row seat to this AI revolution. But I've also witnessed a notable pattern: companies that over-rely on AI-only approaches are experiencing high failure rates, with most purely AI-generated products requiring significant human intervention before launch.

The measurable gap between AI-only product development and human-led approaches with strategic AI assistance is significant—a methodology we've perfected at Perpetual.

The Hidden Cost of AI-Only Product Development

Recent studies reveal a startling reality that AI vendors rarely discuss:

Recently, we witnessed a revealing experiment with a gaming app designed using two approaches: one version built with Replit and another with Lovable. While both games were created from similar AI prompts and shared comparable mechanics (with the Lovable version offering additional features), the crucial difference emerged during our human-led research process.

Through extensive stakeholder and user interviews, we discovered distinct player personas with fundamentally different needs. Some players explicitly stated they "need a challenge," while others preferred games that were "easy to understand but still offered a challenge." Additional research revealed segments who "preferred playing with friends" versus those who "needed something to pass time."

These nuanced user insights—impossible for AI to detect without human research—allowed us to design specific user flows addressing each preference, which were missing from the AI-generated prototypes. When we tested our human-informed prototype, it was an overwhelming success.

Comparison to 2 AI-powered app builders, Replit and Lovable.

Our research-driven approach transformed a generic AI-generated game concept into a targeted experience that delivered what users genuinely wanted and needed, demonstrating the irreplaceable value of human insight in the development process. While AI gave us a head start, validation was key in creating an app that has better chances of market success.

Why Human-Led Approaches Outperform AI

Research consistently demonstrates that human-led teams that strategically leverage AI outperform both traditional approaches and AI-only methods:

1. Problem Framing and Requirements Gathering

Here's where the human-AI gap becomes most apparent - in understanding the unstated needs behind stated requirements:

Where AI Tools Excel:

  • ChatGPT and Claude are remarkably efficient at analyzing existing documentation
  • Perplexity AI excels at consolidating secondary research from multiple sources
  • GitHub Copilot and Cursor can analyze codebases to identify technical dependencies

Where AI Tools Fall Short:

  • They cannot read facial expressions during user interviews
  • They miss cultural context and emotional subtext in user feedback
  • They cannot distinguish between what clients say they want and what they actually mean
"Many organizations underestimated the value of teaming the predictive capabilities of algorithms with the expertise and intuitions of humans, especially in decision-framing."

Our discovery process combines deep stakeholder and user interviews with AI-powered data analysis—a hybrid approach that consistently identifies critical requirements that pure AI approaches miss entirely. AI tools like Dovetail can be very effective when it comes to analysis and synthesis of stakeholder and user interviews.

Dovetail in action providing insights based of compiling different user interviews

2. Design Thinking and Creativity

Even advanced AI tools struggle with the creative leaps that define breakthrough products:

"Human designers using AI tools produced designs with 68% higher originality scores than AI-only outputs, while human teams made "bolder design choices" and AI submissions relied on "standard formats and colors."

Our design team uses AI as an amplifier for creativity—generating alternatives and variations that we then evaluate through a human lens of user empathy by understanding what they hear, what they see, what they feel, and the overall business context.

Screenshot displaying Relume's sitemap feature, which visualizes the hierarchical structure and navigation paths of a website design.
Screenshot showing an example of Relume's wireframe feature being used as a boilerplate to visualize information flow structure.

3. Technical Implementation and Problem-Solving

The most revealing gap appears in handling the unexpected challenges that emerge during development:

86% of GitHub survey respondents report spending AI-saved time on "system design and collaboration"—precisely the areas where humans excel

The Human Excellence + AI Augmentation Advantage

What separates our approach from both traditional development shops and AI-first agencies is the careful balance of human expertise and AI augmentation:

Expert-Led Discovery and Strategy

Our product managers dive deep into your business goals and user needs, creating a foundation no AI can match. We leverage tools like Claude and Perplexity for competitive analysis and market research, but we recognize these are just inputs into a human-driven process.

Part of the magic happens during primary research - when our team sits with actual users, observes their workflows, and builds genuine empathy for their challenges.

A screenshot of how we utilize Claude’s artifacts for data visualization of the competitive analysis.

Design-Development Integration

Unlike AI tools that generate designs disconnected from technical feasibility, our integrated teams create solutions that are beautiful, usable, AND implementable. We use Midjourney and Adobe Firefly to rapidly generate visual concepts and Relume to quickly create wireframes, but human designers make the critical decisions about which elements actually serve user needs. True to our "built to last" philosophy, we create solutions designed for longevity. While GitHub Copilot and similar tools can generate functional code, our senior architects make the strategic decisions about system architecture that determine long-term maintainability.

The Future Is Human + AI, Not AI Alone

As AI tools continue to evolve, the competitive advantage will not come from replacing humans but from creating the optimal collaboration between human expertise and AI capabilities.

"The future workplace will likely be characterized by 'blended teams' where human creativity and judgment work in harmony with AI's computational power and efficiency."

We've been perfecting this balance since before it became trendy. Perpetual’s "built to last" philosophy isn't just about technology choices—it's about creating processes that combine the best of human creativity with the efficiency of AI assistance.

Ready to Experience the Human Difference?

If you've been disappointed by AI-only development approaches or are wondering how to strategically incorporate AI into your product development, let’s talk.

We're not anti-AI—we're pro-results. Our human-led, AI-enhanced approach delivers products that not only launch faster but also succeed in the market and adapt to changing requirements.

Contact us today to discuss how our balanced approach can transform your product design and development journey.