Ayntech

Overview

Ayntech is an enterprise SaaS platform for organizations that inspect physical infrastructure and need to act on AI-detected defects quickly and reliably. Instead of surfacing raw sensor or imagery data, the platform is structured around the way inspectors already think about the work: an inspection contains a set of assets, and each asset carries the defects an AI model has flagged against it — each with a summary, a count, and a way to examine the finding closely before it's reported on.

1. Inspection overview — assets and flagged defect counts at a glance
2. Asset detail — status, type, and defect summary for a single component
3. Defect review — a dual viewer paired with annotation tools to verify the AI's finding
4. Reports — reviewed inspection findings compiled into a deliverable

Problem

A defect flagged by an AI model is only useful if the person reviewing it can quickly confirm — or challenge — what the model found. Without a fast way to compare a flagged finding against the source imagery, an inspector either has to trust the AI blindly or leave the tool to go verify it elsewhere, and the platform's core value gets undermined at the exact moment it matters most. As a multi-tenant product spanning inspections, assets, and defects at three nested levels, the interface also had to hold up as inspection volume grows, not just look clean on a handful of example screens.

Solution

The platform organizes everything around a three-level hierarchy — Inspection → Asset → Defect — keeping each level intentionally light (summaries and counts) until the inspector chooses to go deeper. Opening a specific defect lands directly in a combined dual-viewer and annotation interface, so comparing the AI's finding against the source data and confirming or marking it up happen in one place, not two disconnected steps.

Design decisions

• A nested Inspection → Asset → Defect structure instead of one flat defect list, so every finding stays anchored to the physical component it belongs to
• Dual viewer and annotation combined into one interaction rather than a read-only view plus a separate markup step, so verification happens in place
• A defect summary and count surfaced at the asset level, so the asset list stays scannable regardless of how many issues an individual asset carries
• A single reusable defect-card component used everywhere findings are listed, so the experience stays consistent whether an asset has one flagged issue or many

Outcome

A structure and review flow designed to make AI-detected defects something inspectors can act on with confidence, rather than raw model output they have to take on faith — built on a component system consistent enough to scale across a multi-tenant client base.