Lonere Labs Open Source MIT

Family OS

Open-source scaffold for the chat-first household assistant. Clone it, drop in your keys, rebrand it, ship it in your city — in a weekend.

View on GitHub → Deploy to Vercel

Indian households are quietly drowning in busy work. The errands, the schedules, the staff to coordinate, the bills to pay, the school forms, the doctor's appointments — a daily tax of attention paid mostly by one person, usually a woman.

A real house manager — someone with authority over the calendar, the cook, the school forms, the doctor visits — has always been a privilege of the wealthy. Most families don't have one. The result: capable, ambitious adults losing 40+ hours a month to admin that doesn't need their judgement.

AI changes the unit economics. The same kind of agent that handles a support ticket can handle the maid's leave, the diagnostic test reminder for an ageing parent, the grocery list, the PTA form — at software prices, in the language your grandmother speaks. A house-manager agent isn't science fiction anymore. It's a weekend's work.

The real unlock isn't hours saved. It's productivity unlocked. The mother who was running the house can run her business. The father who was tracking insurance renewals can finish the book. The kid stops being a logistics task and starts being a person. The whole family levels up because attention gets returned to where it belongs.

This won't be a winner-take-all market. There will be hundreds — a house manager for Marathi families in Pune, for elderly parents in Kerala, for joint families in Lucknow, for diaspora families coordinating across timezones. Each needs different cultural texture, a different local network, a different language.

That's why the scaffolding shouldn't be a moat. The moat is your network, your brand, your understanding of your community — not whether you've wired up Postgres and a WhatsApp webhook. So we did that part for you. The bones are here. Take them.

Launch your own. Or build one just for your house.

Nine real skills the assistant already knows about. Each one ships as a small TypeScript object — name, system-prompt hint, examples, suggested schema. Wire one up end-to-end in an afternoon, or leave them as descriptions and let the bot talk about them on day one.

01

Meals & cooking

Plan breakfast, lunch and dinner — respecting dietary preferences and what's in the kitchen.

02

Grocery list

One running list for the family. Recurring essentials. Last-mile to your local chain.

03

Responsibilities

Who does what, this week. Assign, rotate, remind — without nagging.

04

Doctor appointments

Book consults and lab tests, remember follow-ups, surface reports back in chat.

05

Health records vault

Every lab report, prescription and consult note for every member — searchable, queryable.

06

Household staff

Maid, cook, driver, ayah — attendance, salaries, leave, advance, all in one place.

07

School & PTAs

Kids' schedules, exam dates, PTA forms, school holidays — never miss a slip again.

08

Bills & renewals

Electricity, gas, society dues, insurance, subscriptions — paid before they're overdue.

09

Celebrations & gifting

Birthdays, anniversaries, festivals — gifts, sweets, prep, never forgotten.

Channel

Mobile web + WhatsApp

One chat brain, both transports. Add a new channel by writing a single adapter file.

LLM

Any OpenAI-compatible

Moonshot/Kimi, OpenAI, Mistral, Together, Groq. Set a base URL and a model name. Done.

Family

Members & roles

Owner, parent, partner, child, elder, helper. WhatsApp /family invite by phone.

Memory

Postgres + Drizzle

Facts extracted automatically each turn. Next time they message, the assistant knows them.

Voice

Sarvam or Whisper

Voice-note transcription. Indic-native by default. Switch to Whisper for English-strong.

Vision

Lab reports, photos

Plug any OpenAI-compatible vision model for prescriptions, reports, and household images.

$ git clone https://github.com/AbhiK24/family-os
$ cd family-os
$ cp apps/web/.env.example apps/web/.env.local   # add your keys
$ pnpm install
$ pnpm db:migrate
$ pnpm dev                                       # http://localhost:3000