Therness provides AI-powered weld monitoring systems that detect defects like porosity and lack of fusion in real time using thermal and high-speed visual cameras. The platform scores every weld pass/fail during production, generates ISO 17635/3834-aligned evidence packs automatically, and scales across robotic cells and global plants. Its HeatCore AI software runs on-prem via edge computing, with no cloud dependency.
Funding
Funding not disclosed
Founders
Product
Problem
Traditional weld inspection relies on sample-based, post-process methods that are time-consuming, operator-dependent, and prone to missing defects until rework or customer returns. This limited process control delays detection of porosity, lack of fusion, and process drift, increasing scrap and complicating compliance with standards like ISO 3834.
Solution
Therness provides an AI-powered weld monitoring platform that inspects 100% of welds in real time as they are made. The system combines LWIR thermal cameras and high-speed visual imaging (up to 480 fps) to capture the weld pool, while HeatCore AI software scores each seam pass/fail within 100 ms and flags anomalies like porosity, lack of fusion, and process drift. Every weld is automatically logged with thermal and visual frames, timestamps, and process parameters, generating audit-ready evidence packs aligned with ISO 17635, ISO 3834, and ISO 17637. The platform runs on-prem via edge AI, eliminating cloud dependency, and scales across robotic lines and multiple plants through a unified dashboard.
Target Audience
Primary customers are manufacturing plants, welding fabricators, and robotic welding cell operators in industries such as automotive, aerospace, and heavy equipment that need continuous weld quality assurance and ISO compliance documentation.
Features
- HeatCam IR-C and IR-S LWIR thermal cameras with 384×240 resolution, -20°C to 1500°C range, and 60 mK sensitivity for radiometric temperature measurement
- PoolDrop high-speed visual camera capturing Full HD weld-pool video at up to 480 fps with laser-safe arc-glare illumination
- HeatCore AI software with real-time anomaly scoring, pass/fail decisions, and deep-learning defect classification for porosity, lack of fusion, and process drift
- Edge AI processing on-premises with no cloud requirement, ensuring data sovereignty and low-latency decisions under 100 ms
- Automatic generation of ISO 17635/3834-aligned evidence packs including thermal frames, visual frames, verdicts, timestamps, and process parameters
- API integration with QMS/MES systems for seamless export of verdicts and evidence to existing quality workflows
- Three software tiers (Basic, AI, Robotic) supporting recording, statistical process control, deep-learning classification, and pre-integrated robot cell deployment