Every construction project has hold points where specific activities must be verified before the next phase begins. Rebar placement before the concrete pour. Waterproofing membrane coverage before the screed goes down. Formwork inspection before casting. These are not optional checks. They are the moments where getting it wrong means rework at ten times the cost.
The problem is that most teams still verify these hold points through a mix of WhatsApp photos, verbal confirmation, and memory. A photo lands in a group chat, a project manager glances at it between meetings, and work proceeds. There is no reference image to compare against. No structured record. No automatic escalation if something looks wrong. Just a judgment call made on a small screen, often without the context needed to make it well.
With Clappia's no-code platform, you can replace that process entirely. The site engineer uploads both the reference photo and the site completion photo together, submits the form, and the AI Workflow Node compares them automatically after submission. A verdict and brief explanation are written back to the submission record within about a minute. If something does not match, the supervisor gets an instant WhatsApp or email alert. Every check is timestamped, GPS-tagged, and permanently stored without anyone manually writing a report.
Construction apps that organise and store site photos solve a real documentation problem. Photos are searchable, timestamped, and accessible to the project team. That is a genuine improvement over scattered email threads and WhatsApp folders.
But organised storage is not verification. The question "does this photo confirm the activity was completed to the required standard?" still requires a person to retrieve the reference, make the comparison, and log a decision. On a busy site with dozens of activities checked per shift, that comparison step either gets rushed or skipped entirely.
The gap is not about documentation volume. It is about having an automated check that applies the correct standard to every submission, every time, without depending on anyone's availability or memory. The AI Workflow Node enables exactly that. The comparison happens automatically after submission. The verdict is written back to the record. The supervisor is notified instantly if action is needed.
The site engineer opens the app on their phone, fills in the project and milestone details, attaches the reference photo for that activity alongside their site completion photo, adds any remarks, and submits. Once submitted, the New Submission workflow fires in the background. The AI Workflow Node reads both photos and the milestone and activity context from the form, then returns a brief analysis of whether the activity appears complete. The Edit Submission Node writes that result back to the submission. A Condition Node then routes the outcome, alerting the supervisor if something is flagged or notifying the project manager if it passes.
Here is how to build this app from scratch. Sign up for a free Clappia account to get started. The free plan lets you build unlimited apps and access most of Clappia's features to build and test this workflow end to end. Visit the pricing page for a full plan comparison.
Log into your Clappia dashboard and create a new app. Name it something like "Milestone Completion Verification" or "Site Activity Check." This is the central app where site engineers submit verification reports and where the AI verdict lands after the workflow runs.
All blocks in Clappia sit inside sections. The table below lists all the fields needed across three sections. Each field's variable name is what the AI Workflow Node references inside the prompt, so use exactly the variable names listed. See the Labels and Variables guide if you need to edit auto-generated variable names in the right panel.
| Section | Block Type | Label | Variable Name |
|---|---|---|---|
| Site Information | GPS Location | Site Location | gps_location |
| Site Information | Single-Line Text | Project Name | project_name |
| Site Information | Single-Line Text | Region | region |
| Site Information | Single-Line Text | Milestone | milestone |
| Site Information | Single-Line Text | Activity | activity |
| Site Information | Single-Line Text | Site Engineer | site_engineer |
| Site Information | Single-Line Text | Supervisor Name | spv_name |
| Site Information | Single-Line Text | Supervisor Phone | spv_phone |
| Photo Upload | Camera, Images and Files | Reference Photo | reference_image |
| Photo Upload | Camera, Images and Files | Site Completion Photo | site_photo |
| Photo Upload | Multi-Line Text | Site Remarks | site_remarks |
| AI Verification | Multi-Line Text (Read-only) | Verification Result | verification_result |
For the "AI Verification" section, enable the "Display this section if" condition and set it to {verification_result}<>"". This keeps the section hidden while the engineer fills the form and only reveals the verdict after the workflow has written it back to the record. If you maintain a master Activity Register in another Clappia app, use the Get Data from Other Apps block to auto-fill Milestone and Activity from master data instead of free-text entry.
Go to Design App > Configurations > Status and add two custom statuses: "Flagged" for submissions where the AI result suggests the activity may be incomplete, and "Verified" for submissions a project manager has reviewed and confirmed. The default "Submitted" status applies automatically on filing. See the submission status guide for full setup details.
Open the Workflows tab and go to the New Submission Flow. Add the AI Workflow Node as the first node and name it "Analyse Completion Photo." Select your preferred LLM. OpenAI GPT-4o and Anthropic Claude both handle multimodal image comparison well. Clappia provides built-in AI usage credits to get started without your own API key.
In the Instructions field, type @ to open the variable dropdown and select each variable as you reference it. Use this exact prompt:
"Compare the ideal photo {reference_image} with the uploaded photo {site_photo}. These are ideal and actual photos for Milestone: {milestone}, Activity: {activity}. Analyze if the activity is completed based on the photo uploaded. Photos may not match exactly, so try to guess giving some benefit of doubt to the user who uploaded. Final response should be a brief explanation of the decision."
Once saved, the node generates its own variable name, for example ai_node. Note this down for the next step.
Add the Edit Submission Node immediately after the AI Workflow Node. Name it "Write Verification Result." Select your current app as the target. Under Filters, set Submission Id to {$submission_id} to ensure the node always updates the exact submission that triggered the workflow. Under Set Field Values, map the Verification Result field to the AI node's output:
Verification Result {verification_result} → {ai_node}
Add the Condition Node after the Edit Submission Node and name it "Check Completion Status." Instruct the AI prompt to always start its response with "Completed" or "Not Completed" as the first word, making the branch condition simple:
LEFT({ai_node}, 9) = "Completed"
Run a few test submissions first to confirm how your chosen model phrases responses, then adjust the condition if needed.
When the condition is false, fire a WhatsApp notification to {spv_phone}. Include {project_name}, {milestone}, {activity}, {gps_location}, and {ai_node} in the message body so the supervisor gets the full context and AI explanation on their phone immediately. Send an Email to the project management team as a formal documented record of the flagged activity.
When the condition is true, send a routine Email or Mobile Notification to the project manager confirming the activity passed verification.
Use Clappia's Google Sheets/Drive integration to sync every submission as a new row in your milestone completion tracker. Project name, milestone, activity, engineer GPS, AI verdict, and timestamp flow in automatically. The project team gets a live, searchable verification log without needing Clappia access, and the data feeds directly into progress reports or billing milestone claims.
The value compounds across a project. Every submission is structured identically: project, milestone, activity, engineer, GPS, reference photo, site photo, AI verdict, timestamp. When a subcontractor disputes whether their work met specification at a particular hold point, the record answers that question with photographic evidence and an AI comparison. When a client requires quality documentation for a payment milestone, the verification history is already there. When a defect surfaces at a later stage, the records show exactly what was checked and what the AI found.
Patterns in the flagged submissions reveal which activities consistently fail verification, which phases carry the highest rework risk, and where supervision attention needs to increase. That kind of structured insight does not come from a WhatsApp group. It comes from a process where every check is logged the same way, every time, with a defined standard applied automatically.
For teams building wider site management systems on Clappia, this pairs naturally with building inspection apps, daily construction progress reports, and AI-powered quality inspection workflows on the same platform.
Does the AI Workflow Node read both photos automatically once I add the variables?
No, you must reference each image field explicitly in the instructions using @. Both {reference_image} and {site_photo} must appear in the prompt for the comparison to work. Without those references, the AI has nothing to read.
When does the engineer see the verification result?
About one minute after submission. The Verification Result section stays hidden during form filling because of the {verification_result}<>"" display condition. After the workflow writes the result back, the engineer or supervisor can open the submission to read the verdict alongside both photos.
What if the site has limited internet connectivity?
Clappia's offline mode queues the submission locally and syncs it when connectivity is restored. Once it syncs, the New Submission workflow fires automatically and the AI comparison runs. Engineers can fill the form and attach photos without a live connection.
Can one app cover all milestones and activities across a project?
Yes. The Milestone and Activity fields feed directly into the AI prompt on every submission, so the AI always applies the correct context per check. One app handles the entire project lifecycle without any separate configuration per milestone or phase.
Is there a complete audit trail for each verification?
Every submission stores the project name, milestone, activity, GPS coordinates, timestamp, both photos, AI verdict, engineer name, and status. These records sync to Google Sheets/Drive automatically, giving the team a complete, searchable verification log at all times.
Construction quality depends on consistent application of the right standard at every hold point. A photo stored in a folder is documentation. A photo automatically compared against a defined reference, with a verdict written to a permanent record and the supervisor instantly alerted if something is wrong, is verification.
Sign up and start building for free on Clappia. The free plan gives you access to most of the features needed to build and test this complete workflow. Once it is working against your real milestones and site photos, you have a quality verification system running from every engineer's phone, across every project, with no ongoing maintenance required.
Building a broader site management system? See how teams use Clappia for building inspection apps, daily construction progress reports, and AI-powered quality inspection workflows on the same platform.
Get Started – It’s Free
L374, 1st Floor, 5th Main Rd, Sector 6, HSR Layout, Bengaluru, Karnataka 560102, India
3500 S DuPont Hwy, Dover,
Kent 19901, Delaware, USA
.svg){kind=small}
3500 S DuPont Hwy, Dover,
Kent 19901, Delaware, USA
L374, 1st Floor, 5th Main Rd, Sector 6, HSR Layout, Bengaluru, Karnataka 560102, India
.jpg)