FixLine Guardian

A hands-first AI engineer for Quest and glasses: it knows ABB and Siemens machines, every key safety standard and live SCADA. It has multilingual capabilities, and you approve every action.

Comment

Inspiration

Unplanned downtime, avoidable callouts and unsafe repairs cost plants and households real money. Junior engineers and homeowners often have to stop work, look at a manual, and guess. FixLine Guardian is a hands-free mixed-reality plant and home expert that talks you through the repair while you keep your hands on the job — with live SCADA, cited safety standards, project planning, an admin console, and multilingual guidance.

What it does

  • Hands-free guidance: ask by voice or text and get numbered, plain-language steps that explain what to do AND how to do it.
  • Repair & docs: upload manuals, photos and logs for insights; rich repair-type workflows with cited standards.
  • SCADA alarm diagnosis: MQTT / OPC UA / Modbus / MCP plus ISA-style P&ID; turns an alarm into likely causes, checks and fixes; writes need operator approval.
  • Safety first: LOTO/PPE reminders, Safety documents (PTW, JSA, HAZOP, Fire Do/Don’t), USA default jurisdiction with ME/Africa oil options.
  • Field tools: Tuning with live diagrams (shaft alignment), Engineering labs and calculators, Quality visual inspection, Ask AI multi-document copilot with deep-links.
  • Projects: AI PMP/PRINCE2 (and other methods) plans, drag-resize 3xl Gantt, USD↔NGN costing, branded PDF/DOCX deliverables.
  • Multilingual: language bar with IP/geo default, major world languages, and full-page translation when the language changes.
  • Admin console: users, sessions/usage analytics, vendors, machines, a floating Admin + Payments panel, audit — plus a judge-friendly /admin/demo.
  • Mixed-reality modes: immersive WebXR at https://fixline-guardian.vercel.app/xr and a lightweight smart-glasses HUD at https://fixline-guardian.vercel.app/glasses. Main app: https://fixline-guardian.vercel.app
  • Resilient AI: Meta models primary; NVIDIA automatic fallback so guidance stays up.

How we built it

  • Built with IWSDK / WebXR on Next.js, React, TypeScript and three.js, with a futuristic neon/glass UI (Framer Motion), deployed on Vercel.
  • Meta muse-spark is the primary AI model; NVIDIA models are the automatic fallback.
  • Hands-first interaction (hand tracking and voice, no controllers required) and a seated comfort-first design.
  • Purposeful passthrough anchors in XR; clean AR and Smart Glass HUDs keep the optical centre clear (slim top status + bottom dock).
  • Free auth/DB stack, Playwright feature demos, and GitHub → Vercel CI/CD.

Challenges

Keeping answers actionable for beginners, making the model fallback seamless, keeping the Smart Glass HUD readable without obstructing vision, coherent futuristic loading states, and staying comfortable and seated for short task sessions.

Accomplishments

A working end-to-end web app with XR and glasses modes, voice and text guidance, Safety docs and SCADA P&ID, Projects Gantt/PM deliverables, Admin console, multilingual UI, and a resilient Meta→NVIDIA AI pipeline — tested across desktop, tablet and phone widths.

Quantified benefits (illustrative estimates, not measured results)

  • Up to ~30% faster diagnosis of common faults
  • ~20-40% fewer avoidable callouts
  • Potentially tens of thousands of dollars saved per plant per year from reduced downtime
  • Fewer safety mistakes through enforced checklists

What's next

Pilot with a small plant and home-services team in Q1 2027, add more equipment libraries, offline mode and Quest/glasses-native packaging.

Built With

Share this project:

Updates

Submission history