Efficiency Isn't Transformation

Designing AI Products That Actually Change How People Work

How is AI shaping you? We customize models with skills files, preferences, communication styles to shape AI to fit us. But AI is shaping us in the meantime. It’s influencing how we think, how we work, and how organizations operate. Organizations are not designing for that intentionally, and that’s a missed opportunity.

I’ve studied human behavior at every level: the brain, the user, the organization. First as a neuroscientist, then as a product researcher, now as a PM building AI products. That’s why I think about products in terms of human behavior and the systems that shape it.

How Software Products Shape Behavior Today

Most productivity software falls into one of four categories when it comes to influencing how people work. Each has strengths. But none of them capitalize on what AI actually makes possible.

Blank canvas creation. GitHub Copilot, Notion AI, NotebookLM. These tools are neutral amplifiers: fast, flexible, no opinion. The user brings the structure and the judgment. This works beautifully for people who already have a strong process and know what they’re doing. For people who don’t, these tools amplify whatever was already happening, including shallow thinking, bad habits, and mediocre approaches. They help you do more of what you were already doing, faster. If what you were doing was good, great. If it wasn’t, you now produce more of it with more confidence.

Gamification. Streaks, rewards, leaderboards. Duolingo is the canonical example. These mechanics drive retention. Users come back reliably. But what’s actually happening is a dopamine reward loop that reinforces repetition of a fixed action: open the app, complete the task, get the hit. It’s a 1:1 stimulus-response mapping. The user learns to do the same thing consistently, not to think differently or adapt their approach. Gamification makes behaviors repeatable and forms habits, but it doesn’t cause deep learning or change.

Feedback and coaching. Analyze performance, tell the user what to change. Gong and call intelligence platforms are a good example. They record sales conversations, analyze talk patterns, flag missed opportunities, and coach reps on what to do differently next time. But feedback has a fundamental limitation: it’s reactive. It corrects specific errors after the fact, which tends to produce rote updating, not the kind of flexible learning that actually changes how someone approaches new situations. And the timing is hard to get right. After the fact, the moment has passed. During performance, it’s disruptive. Feedback is valuable. But it doesn’t shape how you approach the work in the first place.

Product architecture. Figma says design should be collaborative. Salesforce says sales should flow through pipeline stages. Jira says work should move through tickets and sprints. These products embed strong opinions directly into their architecture, and they genuinely shape how millions of people work every day. This is the closest to real behavior change because the structure itself guides what people do. But it’s rigid. One process for everyone, regardless of context, situation, or user. You either fit the mold or fight it.

Each of these approaches has real value. Blank canvas tools give people speed and flexibility, gamification builds habits, feedback improves accuracy, and static architecture enforces process. But none of them reshape how someone thinks about their work, or ensure the work that gets produced is strategic, high-quality, and actually moves the business.

Challenging the AI Status Quo

AI has the potential to change all of this. But the current hype cycle is brushing some real problems under the rug.

It makes people overconfidently wrong. ChatGPT’s default mode is endless encouragement. Everything you produce is great, every idea is brilliant. That’s designing for engagement, not quality, and it means people are shipping work they haven’t critically evaluated because the AI told them it was good.

Output != productivity. AI allows people to generate an enormous volume of mediocre and shallow output that looks amazing but lacks depth. Documents, decks, prototypes, all polished-looking, produced fast, and mostly not good enough. Organizations are measuring this output as productivity, and nobody is asking whether any of it actually matters.

Workflow automation is not real change. Automating existing workflows with AI delivers real efficiency gains, but efficiency isn’t change. People are still doing the same things they’ve always done, just faster. The potential of AI isn’t to speed up the status quo. It’s to fundamentally reshape how people think about and approach their work.

So How Do You Actually Change Behavior?

So how do we actually do that? It starts with understanding how people learn.

We learn best when we interact with the world through curiosity and exploration. That’s how knowledge gets deeply integrated into how we think and act. We can retain some information by passively reading a document or being told what to do, but it’s not the kind of deep learning that causes lasting change. The question for product builders is: how do you create an environment where people are actively exploring and making choices, but the exploration itself is guided toward high-quality work that moves the business?

Behavioral economists answered half of this decades ago. Thaler and Sunstein’s Nudge demonstrated that people take the path of least resistance, and that you can drive better outcomes by redesigning that path. When school cafeterias rearranged food to put healthier options at eye level, kids ate healthier. No options were removed. No one was told what to eat. The easy choice became the good choice.

I’m combining both of these ideas in how I build AI products. First we create a structured environment that nudges people toward the right path. The reason it produces lasting change, not just compliance, is that within that guided environment, people are still exploring, building, and making choices on their own. That’s deep learning happening in real time. The structure shapes the exploration. The exploration creates changes in behavior.

Now scale that up. Organizations can build their point of view on how value gets created directly into the tools their teams use every day. The AI becomes the way organizations actually get their teams aligned with company strategy, without relying on training programs people forget or playbooks people ignore.

And because it’s AI, the environment isn’t static. The product adapts to who the user is, what they’re trying to accomplish, and what the data says about their situation. The default path is intelligent. The user has freedom to override anything. But the thoughtful path is the easy path, and the system gets smarter over time. This is choice architecture made dynamic.

For sellers, it means the workflow anchors every deliverable in a customer business outcome rather than letting reps default to pitching features. For product teams, it means building what will actually deliver value rather than being a feature factory. The specifics change by function, but the principle is the same: guide people toward the strategic work that actually moves the business.

Here’s where to start: study your 10x employees. The people who consistently do this well. How do they think? How do they approach problems? What workflows do they follow? AI can replicate those workflows and teach the rest of your organization to operate more like your best people. That’s not acceleration. That’s transformation.

Why This Matters Now

The next generation of AI products won’t win on intelligence. They’ll win on having a point of view, and making that point of view dynamic, contextual, and self-improving. They’ll be the products that help organizations stop producing slop at scale and start shaping how their teams think, so the work that gets produced is the work that matters.

The real competitive advantage isn’t AI that produces more. It’s AI that amplifies your best people’s thinking across the entire organization.

---

Donna J. Bridge, PhD is a product leader building AI-native tools for enterprise teams. She holds a PhD in Cognitive Neuroscience from Northwestern University, where she studied the neural mechanisms of learning and memory. She now applies those principles to building AI products that change how people work, not just what they know.