# Doc_Coder

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Syed Kabeer Ali](https://cerebralvalley.ai/u/Doc)
- **GitHub:** https://github.com/MaskedMan-code/PlantPulse
- **Demo video:** https://youtu.be/y_o8vDVIlMo
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4447

PlantPulse is an AI-powered incident investigation assistant designed to help engineers make sense of equipment failures in power plants and other industrial facilities.

The idea came from a simple question: when something goes wrong in a large industrial facility, how do engineers reconstruct what actually happened when the information is scattered across shift logs, equipment reports, maintenance records, and technical documents?

These records may contain different timestamps, incomplete observations, or conflicting accounts of the same incident. Reviewing them manually can be time-consuming, particularly when engineers also need to compare previous failures and identify missing evidence.

I built PlantPulse to make that process more organised, transparent, and evidence-driven.

Users can upload PDF, TXT, and Markdown documents. Using Meta's Muse Spark model, PlantPulse reconstructs an incident timeline, identifies potential contradictions, highlights missing information, compares historical incidents, and prepares a human-reviewable investigation checklist.

One feature I particularly focused on is evidence traceability. Instead of simply presenting AI-generated conclusions, PlantPulse links findings to quotations from the original documents and checks whether those quotations exist in the supplied records. This allows engineers to inspect the evidence rather than relying on the model's interpretation alone.

The application also generates a structured investigation report that can be reviewed and saved as a PDF.

I built the prototype as a solo developer using Next.js, React, TypeScript, Tailwind CSS, Zod, and Meta's Model API, with deployment on Vercel.

For the demonstration, I created fictional industrial incident records containing deliberate inconsistencies, including conflicting pump-trip timestamps and different descriptions of operator actions.

PlantPulse is not intended to replace qualified engineers, determine definitive root causes, or control industrial equipment. Its purpose is to support the human investigation process by making complex records easier to understand, compare, and verify.

My longer-term vision is to develop PlantPulse into a practical investigation workspace for industrial maintenance teams, starting with power generation and potentially expanding to manufacturing and other asset-intensive industries.

Live application: https://plant-pulse-sepia.vercel.app/

## More from Meta Global AI Developer Hackathon

- [Super Ninja](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4444)
- [Deception Check](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4445)
- [Anville](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4446)
- [dotbion](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4448)
- [Granink](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4449)
- [Marksweep (Vanta Point Labs)](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4450)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4447. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt