Feature · AI Weld Inspection
AI-powered weld defect detection — cited to the specific clause
A photo goes in. A formal, standards-cited inspection report comes out — AI-drafted with the specific clause named on every finding, reviewed and signed off by your qualified inspector. No more undocumented judgment calls: every determination carries its citation and its reviewer.
The flow
How it works
- 1
Inspector uploads photo from phone
Any camera works — JPEG, PNG, or WebP up to 10MB.
- 2
AI analyzes against selected standard
ISO 5817, AWS D1.1, JIS Z 3001, or other supported standards.
- 3
Each finding classified by severity
Critical, Major, or Minor — driven by the standard's tolerance bands.
- 4
Each finding cited to the specific clause
Example: ISO 5817:2023 Section 5.2 — Undercut. No ambiguity.
- 5
Corrective action recommended per finding
Repair grind, re-weld, weld-out, or accept-as-is with justification.
- 6
Overall verdict — Pass, Fail, Conditional
Conditional verdicts list every remediation needed to reach Pass.
Welding codes
Standards supported
ISO 5817:2023
Weld quality levels for steel
AWS D1.1:2025
Structural welding — steel
AWS D1.2:2014
Structural welding — aluminum
AWS D1.3:2018
Structural welding — sheet steel
AWS D1.6:2017
Structural welding — stainless steel
ASME Section IX:2023
Pressure vessel and boiler welding
ASME B31.3:2022
Process piping
API 1104:2021
Pipeline welding
JIS Z 3001
Japanese welding standard
EN 1090
European structural steel
CSA W47.1:2019
Canadian welding certification
CSA W59:2018
Canadian welded steel construction
ISO 9606-1:2017
Welder qualification testing
OSHA 29 CFR 1926
Site safety
Tune the engine
AI sensitivity controls
Standard
Balanced detection — routine inspections.
Strict
Higher sensitivity — critical applications.
Lenient
Reduced sensitivity — preliminary review.
Re-analyze capability: change sensitivity without re-uploading the photos. The original images stay attached to the inspection record while the AI produces a new verdict.
Output
What you get
- Inspection findings report with full clause citations
- Digital inspection certificate with QR code (FLG-YYYY-XXXXX numbering)
- One-tap NCR creation from any Critical or Major finding
- Optional countersignature workflow for client sign-off
The numbers
Technical specs
-
Up to 30 photos per inspection
Enterprise plan
-
Analysis time under 60 seconds
End-to-end including upload
-
Works on any phone camera
iOS, Android, modern browsers
-
Supports JPEG, PNG, WebP
Maximum 10MB per photo
Beyond the photo
As-executed weld verification from the machine's own data
Arc Verification ingests what the welding machine actually recorded — amperage, voltage, wire feed speed, travel speed, and derived heat input — and compares it deterministically against the WPS each weld was supposed to follow. An out-of-parameter weld becomes a finding backed by machine evidence; your team decides what happens next.
- Deterministic comparison — every captured parameter is checked against the governing WPS ranges, including Flaggity-derived heat input
- Machine-evidenced findings — an out-of-parameter weld raises a finding with the recorded values attached; a human raises the NCR from it
- Honest capture gaps — what the machine didn't record shows as not captured, what can't be computed as not derivable; gaps are surfaced, never assumed compliant
- Published ingest contract — machine data arrives through the flaggity.arc_data.v1 format, by logger-file import or direct ingest
Ready to see Flaggity on your floor?
14-day free trial. No credit card required. Cancel anytime.