2 minutes; 17 minutes including the linked podcast.
This Advanced SEO Principles and Tactics podcast apprises marketing managers of 2022+ transformations in SEO strategy, and it reminds top SEO pros of how this transformation operates.
That 15-minute podcast is a conversation between two AI personalities generated by DISC’s curated NoteBook LM in response to this prompt:
Help SEO pros grasp the confluence of Google MUM, Entities, and LLM tech in SEO. Focus through (1) the PDF “Overview of SAW (SEO Architecture & Writing)” and DISC’s blog “Now Core SEO Can Long Endure.” Core thesis: Entity and Transformer tech culminate decades of effort to rank sites by what matters to aggregate minds in a given information market. Even the likes of staff conference attendance and customers’ offline interactions influence rank. The “DISC” firm is pronounced “disk.”
[My theories why it failed to pronounce “DISC” are intriguing but not important enough for this post.]
I uploaded to Google’s Notebook LM (NBLM) over 30 top articles on the topic. It summarizes well.
NBLM’s main value is in creating a curated topic library one can query for summaries, nuanced insights, and specific needles in the haystack. Its “Audio Overview” feature generated the podcast conversation using a careful 500-character prompt and a custom conversational style, which proved much better than default settings.
In theory NBLM uses Google’s Gemini LLM tech only on the documents uploaded, but I’ve seen strong evidence that some of Gemini’s pretrained database leaks in. This happened in the above-linked podcast and, more strikingly, in a non-public summary of a court case wherein NBLM mentioned a pattern in the judge’s sentencing that was definitely not indicated in any uploaded documents (and it proved very helpful in resolving that case favorably). I imagine that such “leakage” emerges because NBLM must use Gemini’s codebase to produce comprehensive synthesis. In most cases such use is beneficial, but as always human pros need to vet LLM output. While LLMs can’t reason well yet, their averaging of relevant web-wide content plus curated content, together with well-structured prompts and custom instructions, helps to advance professional understanding. The podcast here shows this synthesis, as does DISC’s NBLM podcast, LLM Reasoning?.
(Both audio files linked above are served from DISC’s Drive, rather than NBLM directly, so that non-Google account holders can listen. Slow internet speeds will require a several seconds to load the audio).
I wrote the post, and no part was written by an LLM.
