Skip to main content

Field Capture

78%
mobile-first field usage
of job-site CRM interactions happen on a phone, not a desktop
photo auto-attach adoption
more field photos logged when auto-attached vs manual upload required
62%
voice notes over text entry
field workers prefer voice-to-text over typing notes on a job site

The mobile app is the primary input device. No forms, no manual entry — the tech taps into the day's project, hits Field Capture, and the camera opens. Photos and videos flow back into the project record automatically. Projects and contacts created from captures on day one of using the app.

Status: Concept · Last Updated: 2026-05-20

The capture flow

A field tech's path from "opening the app this morning" to "this photo is filed against the right job" is four taps. The four screens below trace it left-to-right.

1 · Today's Jobs
7:42● ●
Tue · May 20
Today's Jobs
4 stops · ~6h scheduled
8:00 AMOn Way
445 Maple St
Sarah Garcia
10:30 AMNext
220 Oak Ave
Marcus Patel
1:00 AMSched
77 Cedar Ct
Amy Chen
3:00 AMSched
310 Willow
Greg Foster
↑ Tap a job to open
2 · Job Detail
7:43● ●
Back to Today
Roof Repair
445 Maple St
Sarah Garcia · 8:00 AM
FIELD CAPTURE
Tap to open camera
Call
Notes
Navi
Status
On Way
Arrive
Work
Done
Captures0
No captures yet
3 · Camera
445 Maple St · #4821
PHOTO
VIDEO
BEFORE
● GPS LOCKED
4 · Review · Save
7:48● ●
4 Captures
Review & File
Filing to · 445 Maple St
Before
Before
During
30s vid
0:30
● VOICE NOTE0:42
"Soffit damage on north side, found rot at fascia joint, sending est tonight"
AI · 2 TASKS EXTRACTED
◯ Send revised estimate tonight
◯ Order replacement fascia board

✓ FILE TO PROJECT

Total taps from app open to filed capture: four. Open the app and the day view lands. Tap a job card; the detail screen loads with Field Capture as the largest button on screen. Tap it; the camera opens with the job context already locked in — the photo can't end up on the wrong record. Capture as many shots and videos as the work needs, then tap "File to Project." The voice note transcribes, the AI extracts follow-up tasks, and everything is attached to job #4821 before the tech is off the ladder.


What the camera screen does that a plain camera doesn't

The capture surface looks like any phone camera, but four things are different — each one shaving friction off what would otherwise be a paperwork problem at the end of the day.

Job context locked at the top
The header bar shows the project address and job number. There's no "which project is this for?" prompt later. The capture knows.
Tag-as-you-shoot
A Before / During / After chip floats over the viewfinder. Tap to cycle. Every captured photo lands with its tag already set — no batch-tagging chore at the end of the day.
GPS verification badge
"GPS LOCKED" turns green when the device's coordinates match the job's address within tolerance. If the tech is at the wrong site, the badge turns orange and prompts a re-association before saving.
Photo + video in one mode toggle
The single toggle at the top switches between still photo and short video. Video is auto-trimmed at 60 seconds by default — long enough to narrate scope, short enough not to fill the tenant's R2 with hours of B-roll.
Voice note button after every burst
As soon as the tech taps "File to Project," the review screen prompts "30 seconds of voice notes?" with a one-tap record. Most techs record one; the AI extracts tasks and notes from it before the save lands.
No gallery upload step
The user never leaves the app to deal with the camera roll. There's no "choose photos to upload" prompt later. Every capture goes straight into the project's R2 prefix the moment it's taken — uploaded eagerly when online, queued locally when offline.

Photo Capture → Project Auto-Creation

  1. Contractor takes a photo at the job site — No setup required. Just open the app and shoot.
  2. EXIF GPS coordinates extracted — Every photo has GPS metadata baked in. The app reads latitude/longitude from the file.
  3. Address resolved via reverse geocode — GPS coordinates converted to a street address automatically.
  4. Contact matched or created — System checks if a contact already exists at that address. If yes, attach to them. If no, create a new contact with the address as the identifier.
  5. Project created and photos attached — New project linked to: the contact, the address, the photos, and a timestamp. All automatic — nothing typed.

Video Capture → Tasks and Notes

  1. Contractor records a video at the job site — Narrates conditions, scope, issues, observations — out loud, naturally.
  2. AI transcribes the audio — Full speech-to-text transcription of everything said in the video.
  3. AI extracts structured data — Tasks ("we need to replace the flashing on the north side"), notes, materials mentioned, quantities, and anything else actionable.
  4. Attached to the project — Transcription, tasks, and notes all land on the project timeline with a timestamp.

Digital Receptionist Integration

Sulla Mobile's digital receptionist is embedded in the field app. It handles inbound calls while the contractor is on-site and logs call summaries directly to the project or contact timeline.

Media Storage & Tagging

Input TypeWhat AI DoesTag Applied
PhotoGPS → address → project linkinspection / progress / completion
VideoTranscribe audio → extract tasks/notesfield-video
Audio (standalone)Transcribe → notesvoice-note
Text noteStored as-isnote

Open Questions

  • What happens when GPS is unavailable or inaccurate indoors?
  • Can the contractor manually assign a photo to a different project if the GPS match is wrong?
  • How are multiple photos at the same address across different visits grouped — always same project, or new project each visit?
  • What transcription service — Whisper, Deepgram, other?