Back to home

Stride

AI-Powered Focus Companion

Role: Product + WebStack: Next.js, Supabase, Claudetrystrideai.com ↗

Overview

In 2004, researchers at UC Irvine found that people spent an average of about two and a half minutes on a screen before switching to something else. By 2012, that dropped to 75 seconds. By 2016, it was down to just 47 seconds, with half of all observations lasting under 40 seconds.

Research by Dr. Gloria Mark also found that after an interruption, it takes people around 23 minutes to fully return to their original task.

Most productivity tools respond only after you've already lost focus. They block distracting websites, start countdown timers, or encourage you to begin another session once you've been interrupted. They're useful, but they're all reacting to something that has already happened.

What interested me was the period before someone loses focus.

People rarely become distracted instantly. There's usually a gradual decline in attention, reflected in small patterns throughout a work session. If those patterns can be recognized early enough, there's an opportunity to intervene before focus is lost.

That idea became Stride.

What Stride Is

Stride is an AI-powered focus companion that learns how you work, predicts when your attention is beginning to fade, and sends a timely reminder before you're likely to become distracted.

You start a focus session and work as you normally would. During the session, Stride compares your activity against patterns it has learned over time. If it predicts that your attention is about to drop, it generates a short, contextual nudge designed to help you stay on task.

Rather than interrupting constantly, Stride aims to intervene only when it's likely to be useful.

trystrideai.com
Focus session
03:4611:14 remaining
Currently focused on

Reading a book on software engineering

End session

The biggest difference between Stride and traditional focus apps is timing. Most tools react after you've already become distracted. Stride is designed to step in before that happens.

Features

Predictive Interventions

This is the core feature of the product. As a session runs, Stride continuously compares the current session against your learned focus patterns. When it detects that your attention is likely to drop soon, it generates a contextual reminder instead of waiting until you've already lost focus.

Each intervention is generated with AI using the current session's context rather than selecting from a fixed list of reminders. The goal is to make every nudge feel relevant to what you're doing.

trystrideai.com
StrideFocus check

3 down. Settling in?

Not nowGot it

Focus Fingerprint Premium

Over time, Stride builds a profile of how you typically focus. It learns things like how long your strongest focus periods last, when your attention usually begins to fade, and the patterns that often appear before a drop in concentration.

Premium users receive interventions based on this personal history instead of general heuristics, making the reminders more accurate as more sessions are completed.

trystrideai.com
Your focus fingerprint
90avg focus+3%
Best: Tuesdays
SMTWTFS

Focus Sessions

Everything in Stride revolves around focus sessions. Start a session, work until you're finished, then review it later through your dashboard.

The free and premium plans share the same workflow so new users can experience the core product before deciding whether to upgrade.

  • Free: Up to 3 sessions per day, maximum session length of 45 minutes, general AI-generated interventions.
  • Premium: Unlimited sessions, sessions up to 180 minutes, personalized interventions based on your Focus Fingerprint.

The free plan is intentionally generous. Forty-five minutes is enough time for someone to understand how predictive interventions fit into their workflow.

Dashboard & Focus Trends

The dashboard provides an overview of your focus history without turning productivity into a game. You can review completed sessions, see how your focus patterns are changing over time, and understand when interventions occurred throughout your work.

Session history is displayed using your local timezone so daily activity is always grouped correctly.

Pause & Resume Premium

Focus sessions don't always happen without interruptions. Premium users can pause a session and continue later without losing progress. While a session is paused, Stride also suspends any scheduled interventions so reminders aren't sent while you're away.

Although this seems like a simple feature, keeping session timing and intervention scheduling in sync became one of the more interesting engineering challenges during development.

Core Concepts

Everything in Stride is built around five core ideas. Keeping these concepts simple made it much easier to design the product and make technical decisions throughout development.

Focus Session
The foundation of the product. Every session has a clear start and end.
Predictive Intervention
Estimates when attention fades and sends a reminder before it happens.
Focus Fingerprint
Learns how long you stay focused and the signals that appear before a distraction.
AI Generated Nudges
Uses session context and your fingerprint to make reminders relevant.
Free vs Premium
Based on personalization. Free users experience the core workflow; Premium unlocks deeper personalization.

Architecture

The technology stack was chosen with one goal in mind: move quickly without creating unnecessary complexity. As a solo developer, I wanted managed services wherever possible so I could spend more time building the product and less time maintaining infrastructure.

Next.js 14 + TypeScript: The application is built with the App Router. Server Components handle data-heavy pages like the dashboard, while Client Components power interactive features such as the live session timer. Using TypeScript helped reduce avoidable bugs across the stack.
Supabase: Powers the backend, including authentication, PostgreSQL, realtime updates, and Row Level Security. Using a single managed platform simplified development and reduced operational overhead.
Claude API: Every intervention is generated using Claude rather than selected from a static collection. Each prompt includes session info and the Focus Fingerprint, allowing reminders to be tailored to the individual.
Opik SDK: Because interventions are generated by an LLM, understanding their quality is critical. I use Opik to trace, evaluate, and monitor every intervention, making it much easier to improve quality over time.
Heuristic Prediction: The first version uses heuristics instead of a machine learning model. A lightweight prediction system allowed me to validate the core idea quickly while keeping the architecture simple.
Lemon Squeezy: Acts as the Merchant of Record, removing the complexity of managing international tax compliance and payment infrastructure.

Design Decisions

Designing a focus application comes with an unusual challenge: the interface itself shouldn't become a distraction. Every screen was designed to stay out of the user's way while still providing the information they need.

  • Supportive Interventions. The reminders are written to encourage users to return to their work without making them feel guilty for losing focus.
  • Intervene Before the Break. The product emphasizes prevention rather than reflection. Stride focuses on helping users avoid distractions in the first place, influencing both behavior and the UI.
  • Minimal Session Interface. During a session, the interface stays intentionally simple. Only the information needed for the current session is visible.
  • Clear Premium Features.Premium functionality is clearly identified throughout the application, highlighting where personalization adds value so users understand what they're upgrading for.
  • Straightforward Copy. Rather than relying on marketing language, the interface focuses on clearly explaining what each feature does and setting realistic expectations.