10 Mental Models I Rely On for Building at Scale

Discover how mental models can transform your approach to building software and teams, with 10 powerful models that drive better decision-making and results.

10 Mental Models I Rely On for Building at Scale
"Mental models are deeply held internal images of how the world works, images that limit us to familiar ways of thinking and acting. Very often, we are not consciously aware of our mental models or the effects they have on our behavior."

— Peter M. Senge, The Fifth Discipline

When I joined AWS in 2021, one of the first items in my training queue was an online course called Mechanisms. It started by showing Jeff Bezos describing how "Good intentions don't work. Mechanisms do." Later, the course highlighted mental models as one of the most powerful levers behind effective mechanisms. These models shape the outcomes you aim to produce.

I had never formally identified these models, but a review of my notes revealed that they were already present in my thinking on software, teams, and life. Today, I explicitly document them and pair each with one or more mechanisms: self-reinforcing systems that turn aspiration into habit. These are the top 10 mental models I rely on. I use them whether I'm building software, companies, or myself. So far, I have documented about 30 more and counting.

Top 10 Mental Models

  1. Continuous Learning: Stay curious. The best people I know never stop learning. It's how they stay sharp, relevant, and resilient.
  2. APS Framework (Automation, Pipeline, Security): If you want to move fast and stay safe, you need automation, streamlined pipelines, and security built in from the start—not bolted on later.
  3. Deep Work: Block distractions. Focus hard. Do fewer things, better. That’s how real progress gets made.
  4. Continuous Everything: Anything worth doing should be repeatable. Build systems that keep improving without manual intervention.
  5. Sharing: If it’s worth knowing, it’s worth sharing. Don’t hoard knowledge that might be valuable to others. Externalizing what you’ve learned multiplies your impact.
  6. Feedback Loops: Make it easy to spot what’s working—and what’s not. Use both positive and negative feedback to steer early and often.
  7. Value Stream Thinking: Map the path from idea to customer. Cut what doesn’t add value. Optimize what does.
  8. Growth Mindset: Talent is a starting point. Everything else is effort, reflection, and repetition.
  9. Mechanisms: Good intentions aren’t enough. Set up repeatable processes that drive consistent outcomes. Build systems that improve themselves.
  10. Kaizen: Don’t wait for big breakthroughs. Improve a little every day. Compound gains over time.

A Startup MVP Pattern

Suppose a three-person startup has six weeks to launch its first paid feature—an event-driven SaaS module that enriches raw IoT data and streams results to customer dashboards. Here’s how the ten mental models might line up:

  1. Continuous Learning: Day one starts with a deep dive into the IoT spec and Kinesis best practices. Every morning begins with a 30-minute "learning stand-up" where each person shares a key takeaway. Learning is treated as a daily operating habit—not a side activity.
  2. APS Framework (Automation, Pipeline, Security): Infrastructure is modeled as code across three layers: IaC and automated tests, GitHub Actions pipelines, and security guardrails. These include WAFs, least-privilege IAM roles, and org-level Service Control Policies (SCPs). Every commit reinforces secure, repeatable delivery.
  3. Deep Work: Each founder blocks two hours a day for uninterrupted focus. Slack is set to Do Not Disturb. Architecture diagrams, code, and clear thinking fill the time. The team also holds a daily collaboration session for deep dives—design reviews, pairing, and solving hard problems together.
  4. Continuous Everything: Every code push runs unit, integration, and policy tests. A green build triggers auto-deploy to a preview environment. A feature isn’t marked “done” until its preview URL behaves like production. This makes quality visible and delivery frictionless.
  5. Sharing: Sprint demos are recorded using Loom and shared in the company’s #build-in-public Slack channel. Feedback from early users is encouraged. Transparency beats polish—and every demo is a checkpoint for learning.
  6. Feedback Loops: CloudWatch custom metrics stream into real-time dashboards. If any metric exceeds 2× its baseline, PagerDuty fires an alert within five minutes. These loops surface issues before customers do—minimizing mean time to detect and resolve.
  7. Value Stream Thinking: The deployment pipeline is the value stream—Idea → Code → Build → Deploy → Value—made visible. The pipeline dashboard flags any stage that stalls for over a day in red, prompting daily team conversations about what’s blocking flow—even in a fully remote setup.
  8. Growth Mindset: Code reviews begin by highlighting what was learned (“Clever use of Kinesis partition keys”) before offering suggestions. Failure drills are run regularly—blameless and grounded in data. Improvement is expected, not exceptional.
  9. Mechanisms (Tool): Everything above is codified in a ./bootstrap.sh script. Need a new microservice? Run the script. It spins up tests, pipelines, policies, and scaffolding in under three minutes. Consistency isn't a checklist—it's baked into the tooling.
  10. Kaizen: Every Friday, each team member ships a sub-two-hour improvement: a tighter CloudFormation lint rule, an updated README diagram, or a faster test harness. The improvements are small, but they stack weekly—making the system a little better every week.

How to Start

  1. Write Them Down. Visibility drives action.
  2. Connect Models and Mechanisms. Attach at least one mechanism to each mental model—and at least one mental model to each mechanism. Automation beats aspiration.
  3. Measure What Customers Notice. Deploys per day, escaped defects, P99 latency.
  4. Build Kaizen Into the Culture. Improvements are part of daily work, not side projects. Instituting pre-launch Correction of Errors (CoE) ceremonies ensures systems are production-ready and teams learn before incidents occur.
  5. Teach the Language. When teammates can say, “My mental model is...,” shared understanding scales.

Mental models guide decisions; mechanisms make those decisions stick. Kaizen, now elevated as the tenth model, keeps the whole system improving—especially when no one is looking.

What are your mental models? Write them down, build a mechanism, and share the results. Continuous improvement starts there.