I turned my housing society's WhatsApp group into memory with Cognee
Описание и применение
Автор преобразовал экспорт чата WhatsApp своего жилого комплекса в память Cognee, создав граф знаний, позволяющий быстро получать точные ответы с датой и отправителем, при этом маскируя номера телефонов.
Задача
A large WhatsApp group chat (164 members, 3,320 messages over 239 days) made it hard to retrieve specific information such as water supply schedules because messages were buried and required scrolling.
Результат
The first sync ingested 239 days in ~14 minutes; a repeated import took 15 seconds with no changes; deleting a day emitted a _deleted row and reduced documents from 239 to 238. Queries returned precise, dated answers with masked senders.
Рабочий процесс
- Export chat from WhatsApp (Android or iPhone) to a .txt or .zip file.
- Use cognee.remember with the whatsapp_source connector to ingest the export into Cognee.
- Query the knowledge graph with cognee.search specifying the dataset and node set for the desired group.
- Receive answers that include the date, sender (masked phone numbers), and message content.
- Re-import new exports; the connector detects unchanged days and only processes new or modified days, emitting _deleted rows for days fully removed.
- Mask phone numbers by default to protect privacy.
- Run tests on synthetic exports to validate parsing logic.
Доказательства
My housing society has a WhatsApp group with 164 members. It's where the notices go out: water supply timings, maintenance work, decisions from meetings.
The export I took had 3,320 messages spread over 239 days, and when I wanted to know something simple, like when the water was going to be off, my only option was to scroll and hope.
Cognee works in three steps: you add data, cognify turns it into a knowledge graph, and search asks it questions.
On my society's export, questions like this came back with dated answers, for example "water supply for one hour on August 8", instead of a vague summary of the whole chat.
The first sync took about 14 minutes and the second took 15 seconds, because there was nothing to do.
- Дата записи: 2026-11-01T13:29:12.109Z