ConstruLog uses AI to turn jobsite photos and voice notes into findings and written reports. This page explains what it analyzes, which models run it, what leaves your account to do that, and how much of it each plan includes — with the numbers read straight from the settings the app runs on.
The AI drafts. You sign.
Everything the AI writes is a draft you can edit before it goes out to anyone.
What it does
Each photo — or one frame of a video — gets a description plus safety and quality findings, each with a severity and a recommended action. It also rates image quality and its own confidence, so a blurry shot says so instead of guessing.
Voice notes recorded on site are transcribed to text. From a walkthrough transcript, the AI can also pull out the action items as an inspection checklist.
Models
All AI features run on OpenAI’s API. Each task uses the model that fits it:
| Task | Model | What it does |
|---|---|---|
| Photo analysis | gpt-6.1-sol | Reads photos and video frames at high detail to describe the work and flag safety and quality findings. |
| Reports and daily logs | gpt-6.1-sol | Writes the report narrative and the daily log from the evidence you selected. |
| Voice to checklist |
Your data
Only what a task needs leaves ConstruLog, and only when that AI task runs.
Limits
Each plan includes a monthly amount of AI work that resets every month, on yearly billing too. These are the same numbers the app enforces.
| Plan | AI reports / month | Photo analyses / month | Voice-note minutes / month |
|---|---|---|---|
| Free trial (14 days, total) | AI reports, total15 | Photo analyses, total150 | Voice-note minutes, total60 |
| Free |
Human review
AI output is generated automatically and can be wrong. ConstruLog treats it as a first draft for the person responsible for the work — never as the final word.
Photo analysis uses gpt-6.1-sol, reports and daily logs use gpt-6.1-sol, voice-to-checklist uses gpt-6-luna and transcription uses gpt-transcribe, all through OpenAI’s API. This page reads the same configuration the app runs on, so it always shows the current models.
No. Under OpenAI’s API data-usage policy, data sent through the API is not used to train its models unless we opt in, and we do not.
Only what the task needs: the photo or a single video frame, voice-note audio for transcription, and the project text the report is built from (project name and description, photo and video file names, addresses and capture times, comments, checklist items, transcripts and document names). Never your password, payment details, email address or document contents.
Every account starts with a 14-day trial with full access. No credit card required.
When you build a report, the AI combines the photos, comments, transcripts and checklists you selected into one narrative: work observed, safety, quality, risks, pending items and recommendations.
The daily-log assistant summarizes a day of photos, voice notes and team comments into a short work log, adding the recorded weather for the site when the photos have a location.
gpt-6-luna| Turns a walkthrough transcript into checklist items. |
| Transcription | gpt-transcribe | Converts voice-note audio to text. |
|---|
This table is generated from the same configuration the app calls OpenAI with, so it always shows the models in use. We switch models when a better one ships.
Under OpenAI’s API data-usage policy, data sent through the API is not used to train OpenAI’s models unless we opt in — and we do not.
Read section 4 of the privacy policy| AI reports / month3 |
| Photo analyses / month20 |
| Voice-note minutes / month10 |
| Solo | AI reports / month30 | Photo analyses / month300 | Voice-note minutes / month120 |
|---|
| Pro | AI reports / month90 | Photo analyses / month600 | Voice-note minutes / month300 |
|---|
| Teams | AI reports / month150 | Photo analyses / month900 | Voice-note minutes / month400 |
|---|
| Enterprise | AI reports / monthCustom | Photo analyses / monthCustom | Voice-note minutes / monthCustom |
|---|
Retries are included: regenerating a report up to 2 times, re-analyzing a photo up to 3 times and re-transcribing a voice note up to 2 times do not count against your limits. Voice minutes are counted per note, rounded up. A daily log counts as one AI report.
Compare plans and pricesYes. Photo descriptions, findings and every section of a report are editable. The AI writes a draft; you decide what goes out under your name.
Each analysis includes an image-quality rating and a confidence level, and reports list caveats about unclear evidence. The AI is told to describe only what is clearly visible rather than guess.
Each plan includes a monthly number of AI reports, photo analyses and voice-note minutes, listed in the table above. The 14-day free trial includes 15 AI reports, 150 photo analyses and 60 voice-note minutes.