SiteWire
Every conversation in the right project.
AI & integrations
The problem
On a construction project, the important detail may be spoken in a voice note, photographed on a delivery slip or buried in a WhatsApp group covering several jobs. A search system has to understand the content, assign useful project context and still respect what each worker is allowed to see.
Automated filing must also coexist with human corrections. Reprocessing a message should not silently undo the team's decision about where it belongs.
The solution
SiteWire captures the company's WhatsApp communication and processes messages, documents, photographs and voice notes into searchable project information. Classification identifies the relevant job and document type, while extracted text, dates and references make material easier to retrieve.
Search spans those records alongside tasks, people and internal chat, bounded by project access. Human corrections take precedence over later automatic filing. The application is designed for one company per deployment, with internal discussion kept distinct from the mirrored client conversations.
The decisions behind it
- Search the contents, not just the filename
- A photographed invoice and a spoken instruction need text extraction or transcription before their useful details can be found.
- Human corrections persist
- AI-assisted filing can be revised by the team. A later automatic pass must preserve that correction.
- Context does not override access
- A relevant search result still needs to belong to a project the person is allowed to see.
- Projects, tasks and crew
- Message and document classification
- Search across project communication