Search a focused document set
Import text-based PDF, TXT and Markdown. Choose the sources for each search and retrieve relevant passages from that document set.
A document assistant for Vietnamese SMT engineers. Find relevant technical passages, open the original source, and review troubleshooting evidence before taking action.
Explore the prototype ↗SMT engineering / Technical documents / Human review
Retrieval works in the local prototype. Live AI troubleshooting is under development.
Import text-based PDF, TXT and Markdown. Choose the sources for each search and retrieve relevant passages from that document set.
Each result retains its source document and page or line location. Engineers can inspect the original passage before relying on it.
The Demo mode illustrates symptom analysis and exports a report from the current result. These sample analyses are simulated; the live AI layer is being implemented.
Actual screenshots from the independently checked local M2 prototype, captured on 8 October 2026. The interface is in Vietnamese. The documents shown are synthetic test fixtures, not factory data or official machine manuals.


Demo analysis is separate from document retrieval. It illustrates the report format using simulated output.
Download a sample Demo report (.md) ↗M2 was independently accepted on synthetic fixtures: import, selected-source search, source locations and persistence.
Simulated troubleshooting output and snapshot report export demonstrate the review flow. They are not live AI results.
M3 adds AI analysis and evidence validation. Live AI acceptance and Claude API integration have not yet been confirmed.
Early-stage research prototype. Scanned PDFs without selectable text are not supported yet. Factory-data validation and production release are pending. Engineers must verify recommendations against approved procedures and actual line conditions.
NamTech Agents is an independent project founded by Dinh Van Nam in Vietnam. Our first product, NamTech SMT Assistant, focuses on technical document retrieval and troubleshooting support for SMT engineers.
We start with a practical problem: finding the right evidence in technical documents while investigating a production symptom. Our approach keeps source material visible and the engineer responsible for the final decision.
The application currently runs locally. This website presents the prototype; it does not provide public access to the assistant.
Ask about the prototype, share a workflow challenge, or discuss a demonstration.