SecondBrain
Ask the Brain
Index/Sourceupdated Sat Jun 06 2026 08:00:00 GMT+0800 (Philippine Standard Time)

Sunil's Second Brain Email to IT LT (2026-06-06)

second-brainllm-wiki-patternopenclawmanila-itbrand-fodderp&gexecutive-communicationtestimonial

Sunil's Second Brain Email to IT LT (2026-06-06)

The user's follow-up email to the entire P&G IT LT, two months after his original OpenClaw Personal Learnings note. Captured via Telegram (msg #3234). The email is a working demo of the LLM Wiki Pattern this very vault implements — written from inside the system it describes.

The pitch

"The volume of AI related noise hitting us every week is impossible to filter and to hold in one head. How can we use AI Agents to help us stay on top of what matters and speed up our understanding?"

His answer: a personal Second Brain inspired by Karpathy's LLM Wiki gist (see Andrej Karpathy), running locally on Claude Code + Obsidian, wired to the same OpenClaw agents from his prior note.

The three ingredients (Sunil's framing)

Verbatim from the email — worth preserving as a candidate concept page if it recurs:

  1. Capture anywhere. Share via Telegram, an article through the Chrome browser clipper, or a recording through a Whisper agent. "Two seconds of effort."
  2. Let the agent do the librarian work. Twice a day an agent running with Claude Code reads what was dropped in the raw folder, writes it up, cross-links to everything related, and saves it to the personal Obsidian vault on the PC.
  3. Query own thinking. "When something is on my mind, I ask the vault, not a blank search bar. The answer cites my own notes, and it compounds over time."

The hierarchy this collapses: Listen/Read/Watch >>> Thinking >> Understanding. AI moves consumption to the easiest step; the system moves the work back to thinking and understanding.

The two real-world examples he led with

Example 1 — PHx keynotes → next-morning LT brief

He attended GCC Philippines Summit 2026 (PHx) (McKinsey + Everest, 227 leaders). Both keynote recordings went straight into the system. By the next morning, he had the key takeaways for his Manila LT. The verbatim summary he quotes is the Headcount-to-Value Pivot thesis:

"GCCs are moving from headcount-led to value-led, execution to judgment, efficiency to business impact. McKinsey — PH GCC headcount plateaus by 2030 while value per person rises. Everest — the market still grows ~30% to 2030 net of AI, and you fix process first, technology last."

Compare against the actual vault source: GCC Philippines Summit 2026 (PHx). The summary in the email is faithful to the source page — meaning the system delivered a quotable executive takeaway from raw conference audio in <24 hours. This is the working example of PTSD Debt + Frontier GCC reaching a busy IT leader's outbound communication in a usable form.

Example 2 — "How to elevate P&G Manila IT to be 10X?"

He asked the vault the literal question and the system "connected that question straight back to cross-linked wikis and answered from my notes." The answer cited three sources:

→ This is Elevating Manila IT — A 10X-but-not-Hustle Point of View — the query page filed back to the vault on 2026-06-02. The email is therefore evidence the LLM Wiki Pattern's "file queries back" discipline pays off: a synthesis written four days earlier became reusable brand fodder for an LT-wide email.

What the email asks the IT LT

Two open prompts to peers:

  1. "What would it take for your team to ask its own collective notes a question, instead of starting from zero every time?"
  2. "What would a 10X learning system look like and what outcomes would prove it is working?"

Next steps (his own): figure out how to bring this learning system to the AMA IT LT and Manila Site.

Why this matters in the vault

  • Self-referential validation. The vault now contains evidence of its own enterprise utility — the same way Boris Cherny said he hadn't typed code in 2026, Sunil is demonstrating he is no longer reading raw — he is querying the vault.
  • Brand fodder. A near-ready Medium/LinkedIn spin-off — "How I gave my IT LT a 10X learning system, in two seconds of effort per source" — would extend the existing piece queued in Elevating Manila IT — A 10X-but-not-Hustle Point of View and pair naturally with Designing IT Roles for an AI Era (Talent Strategy POV).
  • Schema check. The three-ingredient framing (capture / agent-librarian / query) is exactly the LLM Wiki Pattern three-operations (ingest / file / query) — but renamed for non-technical readers. Worth keeping the "executive-friendly" vocabulary in mind when explaining the vault.

Open follow-ups

  • Run the actual masterclass link he teased in his PS (currently a stub — "let me know I will share link to the masterclass") — does this point to an internal Confluence page, an external course, or is it a draft? Worth a CRM/journal note if peers ask.
  • Track replies — peers' questions become the next batch of "what to ingest next" for this vault.
  • The vault does not yet have a page for the prior OpenClaw Personal Learnings email referenced ("two months ago"). If it shows up in raw/, ingest it as the companion source; for now it remains an external reference.

Cross-links

Source file: Daily Learning 2026-06-06 08-19 #3234.md