From WhatsApp to site records: how a foreman’s message becomes a timesheet
Do WhatsApp messages get lost on site? See how a foreman’s text, voice note and photo become a work record, a timesheet and an evening report with AI.
The most accurate information on a construction site is on the foreman’s phone: how many people worked, which floor is finished, what is needed tomorrow, who climbed the scaffold without a hard hat. The problem is that this information stays in WhatsApp messages, voice notes and photos. While the office pieces together the evening timesheet, it phones foreman after foreman, and the owner only hears the day’s summary the next morning.
In this article we explain how scattered messages from the field can be turned into structured records, and what to watch for along the way. We use our product SahaSesi AI for the examples.
Why doesn’t site information become a record?
Field crews don’t want to learn a new app; the tool they are most comfortable with on their phone is WhatsApp. Filling in forms, navigating menus or remembering passwords doesn’t fit the pace of a construction site. As a result, information is produced at the right time but piles up in the wrong place.
- The attendance timesheet is pieced together in the evening by talking to foremen one by one; half days and overtime get mixed up.
- Material requests are written in a group chat, and it stays unclear who actually placed the order.
- Defects are reported with photos, but there is no record showing they were closed.
- Occupational health and safety (İSG) incidents are passed on verbally; whether they appear in the report is left to chance.
From message to record: five record types
Most messages from the field fall into one of five types. Sorting messages into these types makes the lists the office needs build themselves.
- Work record: progress information such as “Block B, 3rd floor column formwork finished” is slotted into the daily report with its block and floor.
- Timesheet: “Structural crew 12 people, rebar fixers 4 people half day” is processed per person or per subcontractor crew.
- Material request and delivery: “We need 20 m³ of concrete tomorrow” becomes a request; when the delivery-note photo arrives, a delivery record is opened.
- Defect: a defect reported with a photo is opened against the responsible subcontractor and closed with a closing photo.
- Incident: a work accident, a work stoppage or an inspection visit is kept separate; an occupational health and safety incident rises to the top of the report.
What does AI do in this process?
In SahaSesi AI, foremen and tradespeople write to a single WhatsApp number. A voice message is first turned into text; then the AI reads the message, sorts it by type and reads it back to the foreman as “Saved: …”. If something was misunderstood, the foreman corrects it by writing “no, it was 10 people”.
The foreman writes as usual; the only thing that changes is that the message now becomes a record.
AI can make mistakes. That is why messages that can’t be confirmed or processed land in the office’s “awaiting review” list; no message disappears silently. The daily report is locked once the site chief approves it, and later changes leave a trace.
The evening report: the owner’s one page
The day’s records are slotted into the daily report automatically. Every evening a single PDF, divided into sections by site, goes to managers by WhatsApp and email. The report includes the day’s work records, the person-day total, open defects and, if there are any, occupational health and safety incidents. A site that sent no records is flagged separately.
At month end, the timesheet table downloads to Excel by person and by subcontractor, day by day. There is no need to compile a separate file for accounting.
What to know before you start
- No app is installed for the field crew; you add the person’s name and phone number from the dashboard.
- Records, photos and audio files are stored in a data center in Frankfurt (Germany, EU); the AI and speech-to-text services are in the US.
- We recommend that the office reviews timesheet and occupational health and safety records before using them.
- SahaSesi AI is currently being rolled out through a pilot program, together with the first sites.
What changes on the corporate side?
Site records are only a starting point. The Field Tracking & Planning module covers planned field visits and task tracking, and Occupational Health and Safety Incident / Near-Miss Management covers organization-wide tracking of incident and near-miss records; both carry the same discipline into office processes.
If you’d like to receive your site’s evening report over WhatsApp, we can talk through the scope of the pilot program together.
Frequently asked questions
Does the field crew have to install an app?
No. The WhatsApp they already use is enough. You add the person’s name and phone number from the dashboard, and they receive a short welcome message.
Are messages in a WhatsApp group recorded?
No, the bot can’t read group messages. The message is written to the bot directly, or a message in the group is sent to the bot with “Forward”.
What happens if the AI records something wrong?
Every record is read back to the foreman, who can correct it. Records that can’t be confirmed land in the office’s “awaiting review” list.
What if there is no internet on site?
WhatsApp holds the message on the phone and sends it once there is a signal; the record is written to the day the message arrives.
More articles
What is a corporate portal? Bringing scattered HR, field and audit work onto one platform
The leave form is in email, the asset sheet is in a folder, the visitor book is at reception. A corporate portal brings these scattered tasks under one login, one permission model and one notification layer.
October 21, 2026 · 4 min readWho answers the message that arrives at night? An AI appointment assistant for clinics
Customers write most often outside working hours, in their own language and to several businesses at once. The first accurate reply wins. How does an AI appointment assistant give that first reply?
October 21, 2026 · 4 min read