Six years ago I cared for my own mother and planned the costs on paper because paper was the best tool then. Researching pain points for women from the sandwich generation, I read hundreds of caregiver posts on Reddit, Mumsnet, and Facebook groups alongside reports from the ILO and AARP. The same pattern: one sibling does 70–100% of the care, the rest stay away. The caregiver who has been doing everything for two years has no way to renegotiate the arrangement without the conversation sounding like a threat or an accusation. Existing caregiver contract templates describe the relationship between a hired caregiver and a patient. Nothing covers the arrangement between siblings. So I built CareWeight. A wizard that documents who handles what, maps hours to six care categories with occupation-specific US market rates, and produces a printable agreement. Free, no signup, no backend. US rates only right now. The user is a woman 45–60, mobile-first, often filling this out at 11 PM after a day of caregiving. That shaped every decision. Plain HTML, CSS variables, five vanilla JS modules. No React, no build step. HTML templates hold the content, JS clones and fills them. No server. State in localStorage. Share links compress the full plan into a URL fragment via lz-string. Zero bytes leave the browser. Hosted on Netlify: git push deploy from GitHub, free HTTPS, built-in Forms for the feedback channel, a _headers file for Content Security Policy. No infrastructure to manage. PWA with a service worker. The user might be on a weak mobile connection at home. Offline-after-first-load means the wizard does not break mid-session. Installable from the browser on phone or desktop, no app store. CareWeight is live: https://careweight.netlify.app/ (https://careweight.netlify.app/)If you know someone caring for an aging parent, feel free to share.

0

225

Skill translation platform for professionals 50+

0

9

Crossplatform mobile app. Policy advisors and investors track 2,508 capability fields across all economies: complexity scores, growth trends, unrealized potential gaps. Compare countries, watch specific fields, get alerts on capability shifts. Two primary user types with fundamentally different needs from the same dataset.

0

289

AI code generators break on multi-step forms. After 3-4 steps with validation and branching, the AI loses context and starts breaking things it already got right. You spend tokens fixing regressions instead of building. This template prevents it with strict file separation. Each form step is an isolated component. Modify one step, and nothing else breaks. The AI only needs to read one file at a time.

0

254

Natalia Bukhtiiarova

Design engineer for data-heavy, multi-role products