Skip to content
AppDues
Log in
Contact salesSign up

Hack Week 2025: Five Standout Builds and How to Judge a Sprint

A short internal build sprint works best when the output is visible. This page walks through five winning projects from one 2025 hack week, then turns them into a practical way to scope, demo, and triage your own.

What a hack week is actually for

A hack week is a bounded window for work that does not fit the normal roadmap. The point is not a polished launch. It is speed, cross-team collaboration, and a first honest look at an idea. Some results are small fixes; others are experiments that may never ship.

That framing matters when you judge the output. A demo that answers a question — does this integration actually fire? can a screen reader move through this menu? — can be more valuable than a flashy prototype that hides its limits.

The five winning projects

After voting, five projects stood out across categories. Each one solved a concrete problem for the people using the product or building on it.

1. A paywall preview you can zoom, pan, and fit to screen

Previewing a paywall on a small screen was awkward. The fix added zoom, pan, and dynamic resizing to the Paywalls V2 preview, plus a fit-to-screen option that adapts to the workspace. The result feels closer to a design tool: you can inspect exactly what you are building instead of guessing at a cramped thumbnail.

Why it counts: editor tooling is easy to dismiss as internal polish, but a preview that misleads slows every paywall change downstream.

2. Send test events to integrations from the dashboard

Before this change, checking an integration meant triggering a real user action. The project added a way to send test events straight from the dashboard, so a developer can confirm a webhook, a Slack alert, or an Amplitude event without waiting for a live transaction.

Worked example: a developer wires up a webhook, opens the dashboard, sends a test event, and sees whether the destination received it. If nothing arrives, the problem is in the configuration — not in a missing purchase. That single step removes a long detour through test purchases and log hunting.

3. A retrieval-augmented generation layer for the docs

Documentation can be thorough and still hard to search. This project built a retrieval-augmented generation (RAG) system that pulls the most relevant documentation snippets in response to a question. The aim is faster answers for developers and fewer support requests.

A useful distinction: RAG retrieves passages and grounds a response in them. It is not a general chatbot, and its quality depends on the underlying docs being accurate and well structured.

5. Anomaly detection notifications

This team built a system that watches for unexpected spikes or drops and sends alerts automatically. The trigger could be a bug, an App Store issue, or a surge in new subscribers; in each case the value is reacting sooner rather than discovering the problem in a weekly report.

A caution worth stating plainly: an alert is only as good as its threshold. A demo that fires on an obvious spike still needs tuning before it can be trusted in production.

Other notable projects

Beyond the winners, the week produced a spread of practical and experimental work:

  • A native onboarding survey replaced an Intercom-based version, improving design and data handling.
  • Flutter web support made web subscriptions easier without messy workarounds.
  • Reusable App Store configuration let developers carry existing settings into new apps instead of duplicating setup.
  • Dashboard cancel and refund flows were simplified.
  • A custom icon set replaced third-party icons for a more unified design language.
  • A web billing accelerated sandbox drastically shortened renewal times so subscription lifecycles could be tested faster.
  • Grafana Pyroscope profiling was added to surface backend performance bottlenecks.

A scoping checklist for your own sprint

Use these questions before the week starts and again at demo time.

  1. What decision does the demo help someone make? If the answer is vague, the scope is too broad.
  2. Can a single person finish a visible slice in the time available? Narrow beats ambitious.
  3. Is the success criterion observable? "The webhook receives the event" is testable; "integrations feel better" is not.
  4. Does it remove a step from someone's day? The strongest projects here all deleted a manual detour.
  5. What is the honest status after the demo — live, prototype, or parked? Say which, and do not let a demo imply a launch.
  6. Who owns the follow-up? A promising prototype without an owner quietly disappears.

What the week showed

The pattern across these projects is consistent: give people room to experiment, and the output tends to be concrete. Some of the work was already live; other pieces may grow into full features. The useful takeaway is not the list itself but the habit — small, testable improvements, shown honestly, with a clear next step.

Frequently asked questions

What is a hack week?

A fixed period, usually about a week, when a team sets aside normal roadmap work to build experiments and small improvements quickly. The output ranges from live fixes to prototypes that may or may not ship.

How is a RAG docs assistant different from a general chatbot?

A retrieval-augmented generation system first retrieves relevant passages from a document set, then uses them to ground the answer. That makes the response traceable to the docs, but it also means answer quality depends on the docs being accurate and well organized.

Why send test events instead of triggering a real purchase?

A test event isolates the integration from the purchase flow. If the webhook or alert does not arrive, you know the problem is in configuration rather than in a missing transaction, which shortens the debugging loop.

Are anomaly alerts reliable out of the box?

Not automatically. Detection depends on thresholds and baselines, so an alerting system needs tuning against real data before it can be trusted. Treat a demo as proof of concept, not proof of production readiness.