cd /news/artificial-intelligence/proof-of-practice-nobody-sees-the-4-… · home topics artificial-intelligence article
[ARTICLE · art-56544] src=dev.to ↗ pub= topic=artificial-intelligence verified=true sentiment=↑ positive

Proof of Practice-Nobody Sees the 4 AM Practice. Now There's Proof It Happened.

A developer built Proof of Practice, a decentralized ledger that records daily dedication such as coding sessions and study hours. The app uses Google Gemini for analysis, ElevenLabs for voice feedback, and mints milestones on the Solana Devnet blockchain. It is built with Next.js, TypeScript, and Tailwind CSS, integrating Snowflake for session history.

read2 min views1 publishedJul 12, 2026

*This is a submission for *Weekend Challenge: Passion Edition

#

What I Built

Proof of Practice is a decentralized ledger that turns invisible daily dedication into a permanent, verifiable record.

Passion is defined by the effort we put in when no one is watching—the late-night coding sessions, the scale drills, the quiet hours of studying. Too often, this grind disappears without a trace. Proof of Practice makes that work visible. It captures your reflections via voice or text, analyzes your logs using Google Gemini, synthesizes spoken coaching feedback with ElevenLabs, tracks your progress on a streak dashboard, and lets you mint verified learning milestones directly to the Solana Devnet blockchain.

#

Demo

Guest Access: Click "Try Demo Mode" to test instantly with a temporary Solana Devnet keypair generated on the fly (no Phantom extension setup needed).

#

Code

#

How I Built It

Proof of Practice is built using Next.js 15 (App Router), TypeScript, and Tailwind CSS.

The core application workflow coordinates four API-driven services in a single server-side workflow:

Snowflake (The Memory): Every log request retrieves the user's last 5 practice sessions from a Snowflake data warehouse using connection-pooled SQL binds. This gives the AI full context of the user's ongoing learning history. The session details are then saved back to Snowflake to populate the analytics dashboard. #

Google Gemini (The Brain): We use gemini-3.5-flash to read the user's transcription history and logs. It acts as a supportive coach, generating structured JSON feedback that addresses what the student practiced. Gemini also writes a cinematic 40-word narrative script when a milestone is minted. #

ElevenLabs (The Voice): The feedback is converted into natural speech using ElevenLabs' eleven_turbo_v2

model with a standard pre-made voice (Roger). This audio is rendered inside a custom visualizer bubble. We wrapped this call in a try-catch

safety net: if ElevenLabs hits rate limits, the app continues gracefully with the text-only feedback. #

Interactive Frontend Details: The user interface features custom-designed Light/Dark modes, a live voice recording canvas visualizer utilizing the Web Audio API AnalyserNode

, and haptic sounds generated dynamically via HTML5 Web Audio API oscillators.

#

Prize Categories

Best Use of Snowflake (Context-aware retrieval and session persistence) #

Best Use of Solana (On-chain milestone minting via Solana Memo transactions) #

Best Use of ElevenLabs (AI feedback text-to-speech synthesis and narrated trailer voices) #

Best Use of Google AI (Practice analysis and JSON generation with Gemini 3.5 Flash)

── more in #artificial-intelligence 4 stories · sorted by recency
── more on @google gemini 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/proof-of-practice-no…] indexed:0 read:2min 2026-07-12 ·