Presenting agentic AI during the AISG and Micron MOU signing

On 30 January 2026, I presented during the AI Singapore and Micron MOU signing. The project I spoke about was an updated agentic AI productivity tool for yield-related root cause analysis and first-pass wafer issue diagnostics.

I tried to keep the framing away from internal names and closer to the actual problem. Engineers spend time collecting context, checking historical signals, comparing possible causes, and deciding what to investigate next. The agentic workflow was meant to support that process by retrieving relevant information, reasoning through investigation paths, and helping the user reach a useful first hypothesis faster.

The audience was mixed, which made the presentation more challenging in a good way. Some people cared about architecture and model behaviour. Some cared about engineering outcomes. Others mainly wanted to know whether this kind of tool could scale and be adopted beyond a small prototype.

That forced me to be clearer about what the system was doing and what it was not doing. The tool was not replacing engineering judgment. It was helping to organize information and reduce the time needed to get to a starting point for diagnosis.

The global fan-out afterwards was another useful learning point. Adoption is not only an engineering problem. It is also documentation, trust, repeated explanation, and making sure the tool fits existing ways of working. I still find that part difficult sometimes, but it is also the part that makes the work feel real.

Footnote: Ported over from my personal blog. Initially posted on 30 January 2026.