Building AI That Actually Helps

There's something really human starting to happen in AI as people put away the frantic AI popups and chatbot reskins and start to think deeply about the world we live in and what people need from technology. I'm talking about those moments—and I've had a few more of them in the last week than I had in the last three months—where a well-designed AI feature genuinely makes my day a little easier, a decision a little clearer, or a process a little more human.

The earliest days of the mad-dash are starting to mature. Some of the reskins and repackaging of old technologies are falling out of the bottom of the market, and what's emerging feels different. User experience is coming back into AI, and it's going to change things all over again.

There's this sweet spot as someone working in tech—where human needs that we addressed poorly meets new machine capability and changes things for the better. When you find it, the results can be genuinely transformative. That's why I'm starting this newsletter: to share what I'm learning about building AI that actually helps, and to connect with others who are working toward the same goal.

I used to be a consultant for Pivotal Labs, among other places. I've worked on big projects and small ones, but the ones that stand out—the ones that did the most good—were where technology was pointed at a specific purpose and a clear need.

On one project, we were trying to help people working in a call centre deliver better healthcare support. As the designer, I asked: "What if we made the data designed for phone calls?" That simple question became our design principle, but it did a lot of work to help us make smarter decisions more effectively.

We filtered the data specifically to improve these phone calls because they couldn't be automated. Clear numbers, highlights on specific bits that mimicked the paper processes employees had built over the years to make a system that didn't work actually work for them. The system we created helped them focus on problems that needed human creativity while automating the routine patterns they'd been handling for years.

What made it work wasn't the complexity of the technology—the underlying designs looked fairly simple. But how quickly and intuitively people could use them? That was game-changing. What got us there wasn't having the best cloud infrastructure or the most sophisticated algorithms. What made a real difference was the clarity of purpose and deep respect for the people who would use these systems.

That experience taught me something important about finding those sweet spots—and it's made me particularly excited about the moment we're in now.

Right now, we're at this fascinating inflection point. AI capabilities are becoming genuinely accessible to product teams, but the approaches for integrating them thoughtfully? Still being figured out.

This creates an incredible opportunity—not by adopting every new tool that comes along, but by developing sustainable practices that can evolve with the technology. I'm talking about evaluation frameworks that account for model drift, governance structures that scale with team growth, and integration patterns that don't create vendor lock-in.

But here's what I find most exciting: AI's potential to help us build more inclusive, accessible products. Imagine interfaces that automatically adjust complexity for different technical literacy levels, real-time translation that preserves cultural context, or checkout flows that adapt for varying accessibility needs. The question isn't whether we can do it—it's how we can do it responsibly and effectively.

And I've seen what happens when teams get this right. The retail team I'm working with now has built recommendation systems that actually broaden discovery rather than narrowing it. Their AI-powered customer service tools help staff resolve issues faster without making them feel surveilled. These aren't impressive demos—they're sustainable systems that get better over time.

My goal is to create a space where we can have practical conversations about AI in product development. I want to share insights from my current work in retail technology, where we're exploring how AI can create better experiences for millions of customers while supporting sustainability goals. I'll also draw on patterns I've noticed across different industries and lessons learned from both successes and failures along the way.

Here are the topics I'm most excited to explore with you:

Building Sustainable AI Capabilities: How do you create AI-powered features that get better over time rather than becoming technical debt? I've seen recommendation engines start producing increasingly narrow suggestions and personalization features begin feeling creepy rather than helpful. What does it mean to build AI systems that can evolve with your organization's needs?

Human-Centered AI Design: How do you design AI features that feel genuinely helpful rather than intrusive? What does good product management look like when your features include elements you can't fully control—when you're essentially product managing probabilities rather than deterministic outcomes?

Cross-Industry Innovation: What can retail learn from healthcare's approach to AI governance around patient consent? (Spoiler: quite a lot about transparency.) How might telecom's infrastructure thinking help e-commerce platforms build more resilient personalization systems? I love exploring how insights from one domain can spark innovation in another.

Practical Implementation: Beyond the theory, what does it actually look like to evaluate AI tools, integrate them into existing workflows, and measure their impact? How do you build teams that can work effectively with AI capabilities?

I don't pretend to have all the answers—anyone claiming they do in this space probably isn't someone you want to learn from. What I can offer is a practitioner's perspective from someone who's been building products for a while and is genuinely excited about what becomes possible when AI is implemented with intention.

But here's what I'm most excited about: the teams seeing genuine AI success aren't necessarily the ones with the biggest budgets or the most sophisticated models. They're the ones who've figured out how to maintain human agency while amplifying human capability. They're building systems that help rather than replace, that clarify rather than complicate.

I'm genuinely curious about your experiences too—particularly the moments when AI implementation surprised you, either positively or in ways that taught you something unexpected. The best insights often come from the intersection of different experiences and perspectives.

The next few years are going to be fascinating for those of us building products. I genuinely believe we're living through one of those rare moments when individual practitioners can meaningfully shape how an entire field develops. AI capabilities will continue to improve rapidly, but the real innovation will happen in how we integrate those capabilities into experiences that genuinely serve people's needs.

We have an opportunity—and a responsibility—to shape how AI gets embedded into the products and services people use every day. We can choose to use these tools to create more inclusive, more helpful, more sustainable experiences. But only if we approach the work with intention, empathy, and a commitment to getting better over time.

That's the conversation I want to have—one focused on building rather than just adopting, on sustainability rather than just speed, on genuine helpful learnings and ways of thinking rather than sales pitches.

What's your experience been with AI in product development? I'm particularly curious about the successes that genuinely surprised you, the challenges that taught you something unexpected, and the questions you're wrestling with for your next implementation.

Connect with me here or on LinkedIn—I'm always eager to learn from fellow practitioners who believe that building AI that actually helps is both a technical challenge and a fundamentally human one.

Thanks for reading, and expect our first proper post tomorrow.

– Saielle