VariableWay

VariableWay is a small, practice-first organisation that explores composition-first tooling: try new tools, combine AI with existing systems, and build small, useful products that connect them.

We avoid puffery and marketing language. Our work is grounded in short, repeatable case studies, practical recipes, and problem-solving: prototypes you can run, learn from, and reuse.

Core idea

• Everything is a variable you can connect. Treat tools, data sources, and UI components as composable pieces — wire them together to solve concrete problems.

Principles

• Practice-first: experiments must produce a working demo or recipe.
• Composability: prefer small focused pieces that can be re-used and recombined.
• Signal over noise: surface useful information and feeds that accelerate experimentation.

What we build

• Tool playgrounds: compact repos that test an SDK, API, or workflow end-to-end.
• Connectors: small adapters that link two or more services (e.g., RSS → summarizer → Slack digest).
• Mini-products: one-purpose apps that demonstrate value quickly (automation, monitoring, data enrichments).
• Tutorials & recipes: copy-pasteable step-by-step guides with expected outputs.

Example case studies (practical)

• Daily research digest: subscribe to a curated feed, extract key sentences, run an LLM summarizer, and send a concise digest to a chat channel.
• Screen-to-data: capture a snippet (image or text), run an extractor, enrich with external APIs, and append to a searchable personal knowledge store.
• Event-triggered automation: when a high-signal item appears in a feed, run a short pipeline that triages, annotates, and opens a tracked issue for follow-up.

Revisiting daily workflows with AI

• We regularly re-evaluate common daily workflows and show how AI can improve or simplify them. Each entry is a short case study: the problem, a minimal pipeline, and the measured benefit.

Examples

• Email triage: use classifiers + summarizers to surface 5 actionable messages each morning; link to the source and suggested reply drafts.
• Meeting notes → actions: capture meeting transcripts, extract decisions and action items, and push them to a task tracker with links and owners.
• Research deep-dive: turn a set of saved links into a structured brief: key claims, supporting evidence, and follow-ups.

How we document these studies

• Problem statement (1–2 lines)
• Minimal pipeline (tools and connectors)
• Commands to reproduce and sample outputs
• Short reflection: what worked, what failed, and what next

Org activity status

• This repo can maintain a simple profile/ACTIVITY.md that lists recent public events for the organization. A scheduled GitHub Actions workflow (.github/workflows/refresh-activity.yml) runs daily and updates that file using the repository's GITHUB_TOKEN.

Customization

• Change the schedule in refresh-activity.yml (cron) to adjust how often the activity file is refreshed.
• The script uses the org name from github.repository_owner. To target a different org or filter events, modify the curl call or add query logic.

Notes

• The workflow uses the REST API and jq/awk to produce a compact list; you can replace that logic with a richer Node/Python script if you need more structured output or HTML.

Org analytics

• See profile/ANALYTICS.md for a lightweight public analytics panel that uses GitHub Readme Stats and similar services to display contribution stats, top languages, streaks, and trophies for the org.