Yes, vibe-coding is useful.

But the most powerful way to use AI may be deep research.

Ross Morsali is the developer of the Search and Filter plugin. He uses deep research to find entirely new solutions to complex WordPress problems.

For example, WordPress search is normally slow for large sites. If you have over 10,000 WordPress posts, it can take a while for your visitors to see search results.

Ross used deep research to find entirely new solutions to this problem. With deep research, AI can investigate complex questions, crawl many sources, cross-reference claims, and synthesize the findings.

This is a great conversation with a developer who takes his work seriously and loves to break new ground. In the second half of the podcast, Ross talks about taking his deep research to build innovate answers in a new area: speech-to-text.

0:00 — Steve introduces Ross and the deep-research topic.
2:07 — Ross explains the search/filter performance problem.
4:11 — Ross describes indexing and the technical foundation behind fast search.
5:12 — Ross explains how deep research with AI helped him explore new approaches.
6:29 — He breaks down how he uses research mode in Claude.
8:22 — Ross describes how long the research phase took.
9:21 — He explains the bitmap approach that improved performance.
11:23 — Ross talks about using AI coding tools alongside research.
12:30 — The rewritten indexing system shipped in January.
15:42 — Ross shifts to a new rabbit hole: speech-to-text.
17:40 — He explains why low latency is critical for voice input.
19:30 — The voice tool concept expands into a much broader system.
22:17 — Ross says they’re in the polishing stage and still refining details.
27:25 — He talks about training a new model and applying for a grant.
29:57 — Ross explains his privacy-first mindset and why data ownership matters