GlowSteps: Micro Goals Tracker

Meet Glowy: Your daily companion for turning dreams into action

Comment

GlowSteps: Micro Goals Tracker

Inspiration

Gabby's brief was brutal and honest: ambitious women with huge dreams (solo travel, six-figure salaries, financial freedom)… stuck in "save this for later" mode instead of actually doing things.

I didn't want another habit spreadsheet. I wanted an app that says:
"Here's one small thing. Do it. Feel good. Repeat."

So I made GlowSteps: pick one dream, do one micro‑action, and a tiny firefly lights up to celebrate.

What it does

GlowSteps is intentionally simple:

  • You pick one goal (solo travel, salary negotiation, or financial freedom).
  • Every day, you get AI‑generated micro‑actions tailored to that goal.
  • Paid users get a personalized starter pack on first use:
    • Originally 3‑2‑1 format (3×5‑min, 2×10‑min, 1×20‑min),
    • Simplified to 1‑1‑1 (one 5‑min, one 10‑min, one 20‑min) after realizing that generating the full JSON set took too long and slowed the experience.
  • When you tap "Done", your streak increases and Glowy the Firefly glows a bit brighter.

No giant dashboards. Just one clear step and a small win.

How I built it

Stack, in one breath:

  • Rork + Expo (React Native) for the mobile app.
  • TRAE also helping me when my rork token run out.
  • Supabase for auth, database, and Edge Functions.
  • RevenueCat for subscriptions and entitlements.
  • Gemini as the main LLM.

I added a retry layer: if Gemini hits a limit, the app automatically switches to another model. If AI still fails, it quietly falls back to a curated action library, so the user still gets a useful micro‑action instead of an error.

Auth almost killed this project. I'm from a Flutter background, and React Native auth was burning Rork credits while refusing to behave. So I cut everything down to "Login with Google to Continue Onboarding" to reduce friction for users and friction for me.

Challenges I ran into

  1. Auth chaos: React Native auth with google that restart to root '/' route patterns fought me constantly. I eventually simplified to make make login mandatory for better onboarding (since I need to make a personalize action I need to make sure only registered user can use my LLM feature) to save time and Rork credits.

  2. AI generation speed: The original 3‑2‑1 personalized starter pack (6 actions total) took too long to generate as a single JSON response. I scaled it back to 1‑1‑1 (3 actions) so users wouldn't wait 20+ seconds on first launch.

  3. Scope creep vs. deadline: I wanted multiple goals, full Glowy customization, and a rich cosmetic system. Reality check: time + credits said no (working as full-time too). I shipped one goal, one mascot, one tight loop instead of ten half-baked features.

  4. Rork credit management: Every UI iteration burned credits. I had to design carefully, test locally when possible, and cut unnecessary complexity to stay within budget.

Accomplishments that I've proud of

  • AI that actually works: GlowSteps generates unique, personalized actions per user based on their specific goal and gracefully falls back to a curated library if AI fails. No error screens, no broken experience.

  • First tester loved it: I showed the app to my wife. She immediately loved the mascot concept and started imagining outfits and variations for Glowy features that don't even exist yet. That told me the core character and daily loop are already engaging enough to carry the experience.

  • Simple but emotional: The app is intentionally minimal, but tapping "Done" and watching Glowy glow brighter feels surprisingly satisfying. One tight loop beats a feature dump.

  • Shipped on time: Despite auth issues, AI tuning, and credit constraints, I got a working MVP with RevenueCat monetization live for judging.

What I learned

  1. One strong loop beats a feature list: GlowSteps works because the core experience is clear and repeatable: goal → action → streak → glow. Every cut I made strengthened that loop.

  2. A tiny character can make pressing "Done" feel emotional: I underestimated how much personality a simple mascot could add. Glowy isn't just decoration it's the visual reward that makes daily actions feel meaningful.

  3. Cut scope early, cut scope often: Shipping a single-goal, single-mascot MVP is what allowed the AI + core experience to actually finish. I'd rather have one polished feature than five broken ones.

  4. Test with real humans ASAP: Even one early tester (my wife) gave me critical validation that the mascot concept and simplicity were working before I burned more time on features nobody asked for.

What's next for GlowSteps: Micro Goals Tracker

I'm planning to rush character customization this year as the first major milestone post-hackathon:

  • Better onboarding flow: to make user at least feels the tip of main feature instead of forcing to login.
  • Full Glowy customization system: unlock new skins, outfits, and lantern themes tied to streaks and milestones (travel outfit, nighttime look, golden lantern, cozy room, etc.).
  • Multiple goals support: let users pick 2–3 dreams and toggle between them, with unique AI-generated actions for each.
  • Shareable "I did it" cards: one-tap sharing to Instagram Stories or TikTok so users can celebrate wins publicly and bring friends into the app (creator-native distribution).
  • Push notifications: warm, non-guilt daily nudges ("Glowy's waiting for you!" instead of "You missed your streak").
  • Deeper Gabby integration: link specific micro-actions to Gabby's blog posts, workshops, or YouTube content so the app feels like a natural extension of her community.
  • Milestones & badges: visual rewards for 7-day, 30-day, and 100-day streaks, plus special Glowy animations when you hit big goals.

GlowSteps doesn't try to fix your whole life.
It just makes today's step obvious and lets a tiny firefly cheer you on for taking it.

Built With

Share this project:

Updates