A little lab for big ideas.

ARCHIVED EXPERIMENT / 2026-02

News-to-Action Digest: Turn Trending “новости” Into Team-Ready Updates

Build a lightweight internal digest that converts high-volume news queries into actionable, non-political operational updates for product and engineering teams. The feature pulls headlines from approved sources, clusters duplicates, summarizes in plain English, and routes items i

12.0 modeled hours saved

Original illustrative model: 80 baseline hours × 0.6 efficiency × input / 100. This is an experiment, not a measured outcome.

Source: google trends · KZ · Captured 2026-02-18

ARCHIVED EXPERIMENT / 2026-05

Rocket Launch Watch: A Developer-Friendly Space Events Digest

Build a monthly, portfolio-ready feature that turns trending interest in космонавтика (cosmonautics) into a practical product: a lightweight “Space Events Digest” that aggregates upcoming rocket launches, mission milestones, and livestream links, then summarizes them into a clean

1. Source discovery & schemaIdentify 2–3 reliable sources (official agency pages, launch schedule sites, RSS/Atom feeds). Define a normalized event schema: mission_name, provider, vehicle, window_start, window_end, location, liv0.8 weeks
2. Ingestion pipelineImplement fetchers with retries, rate limiting, and HTML-to-structured extraction where needed. Store raw payloads for traceability. Schedule daily runs and emit structured events into a database.1.2 weeks
3. Deduplication & quality rulesAdd canonicalization (time rounding, name normalization), fuzzy matching for mission names, and conflict resolution (prefer official sources). Flag uncertain fields and track provenance per attribute.0.9 weeks
4. Summaries & localizationGenerate short, factual summaries from structured fields (not freeform scraping). Provide English UI copy and allow optional RU/KZ labels. Add guardrails: no speculation, include source links, and sho1 weeks
5. Mobile-first UI + calendar exportBuild a feed with filters (This week / This month / By provider) and an event detail view. Add ICS export and deep links to livestreams. Implement caching and skeleton loading for slow networks.1.1 weeks

Source: google trends · KZ · Captured 2026-05-02

ARCHIVED EXPERIMENT / 2026-06

eGov KZ Task Automator: One-Click Document Checklist + Status Tracker

Build a lightweight web app that helps users and support teams prepare and track common eGov KZ requests (e.g., certificates, registrations) without storing sensitive data. The feature generates a step-by-step checklist, required document list, and a shareable progress link. It a

1. Scope top 5 workflows + content modelSelect the most common eGov KZ tasks for your audience, define required documents/steps per workflow, and design a JSON schema for checklists, prerequisites, and help text.0.8 weeks
2. MVP UI: checklist + progress + share linkImplement a mobile-first checklist with progress indicator, step details, and a shareable read-only link (no PII). Add local-only notes per step.1.2 weeks
3. Copy-ready request summary + templatesGenerate a standardized summary (task, completed steps, missing docs, questions) with one-tap copy. Add templates for support teams to reduce back-and-forth.0.8 weeks
4. Reminders + offline-first hardeningAdd optional reminders (calendar export or local notifications where supported), cache assets for offline use, and ensure graceful degradation on older mobile browsers.0.7 weeks
5. QA, accessibility, and analytics (privacy-safe)Test on common KZ devices/browsers, add keyboard and screen-reader support, and implement privacy-safe event tracking (no user content, no identifiers).0.5 weeks

Source: google trends · KZ · Captured 2026-06-02

ARCHIVED EXPERIMENT / 2026-07

Match-Day Automation Pack: Turn Live Sports Spikes into Reliable Ops Signals

A portfolio-ready feature that helps teams prepare for predictable traffic spikes driven by live events. Using the trend keyword as a proxy for a sudden attention surge, this feature guides an engineer or PM through estimating extra support/ops time, selecting a lightweight mitig

2.5 modeled hours saved

Original illustrative model: 120 baseline hours × 0.35 efficiency × input / 100. This is an experiment, not a measured outcome.

Source: google trends · KZ · Captured 2026-07-02