🧠 Building a Personalized AI Life Coach with Obsidian + Claude

By Aaron Chung · March 2026

This guide captures the exact system I built and the process of building it — not theory, but what actually worked.

Two Ways to Get Started

Not technical? No problem. You don't need Git, GitHub, or a terminal to use this system. Download the vault template below, open it in Obsidian, and start writing. You can set up backups later — or skip them entirely and use iCloud, Dropbox, or Obsidian Sync instead.

⬇️ Download the vault template (ZIP) — no account required, just unzip and open in Obsidian.

If you're comfortable with Git and want automatic version-controlled backups across devices, follow all 9 steps including Step 5. If you're not, skip Step 5 entirely — the rest of the system works just fine without it.

What This System Does

Most people use AI as a search engine with better grammar. You ask a question, it answers, you close the tab and forget it happened.

This system is different. By giving an AI persistent, structured access to your personal knowledge base, you create something that genuinely knows you — your goals, your patterns, your strengths and blind spots, your history. Over time it becomes less like a tool and more like a thinking partner who remembers everything you've ever told them.

The outputs this system has enabled in practice:

A structured self-analysis identifying genuine strengths, weaknesses, and career misalignments with more clarity than years of introspection alone
A career direction note that reframed a job search from reactive to strategic
Targeted company research mapped to personal profile and constraints
Behavioral interview prep grounded in real career history
Study notes with real depth (not surface summaries) linked to prior knowledge
A complete hardware/software inventory with optimization recommendations
Personalized daily and weekly planning that reflects actual priorities

The core mechanic is simple: your vault is your context, and context is everything.

What You Need

Essential:

Obsidian — free, local-first, markdown-based note app
Claude — Claude's desktop app with Cowork mode (folder access) enables the AI to read and write directly to your vault. Any AI tool with local folder read/write access will work.

For backup — choose one:

Non-technical: Store your vault folder in iCloud, Dropbox, or Google Drive — just move the folder there after creating it. Or use Obsidian Sync ($10/month) for seamless multi-device sync built into the app.
Technical: Git + a GitHub account — free, gives you full version history and cross-device sync via terminal. See Step 5 for setup instructions.

Recommended Obsidian plugins (all free):

Obsidian Git — auto-commits and pushes your vault to GitHub on a schedule (technical users only)
Dataview — query your notes like a database (powers the "notes created/visited today" sections)
Templater — smarter templates with dynamic date variables
Calendar — visual calendar sidebar linked to daily notes

Helpful but optional:

Raycast (Mac) — launch Obsidian notes by name, clipboard history, snippets
NotebookLM — good for synthesizing large documents from your vault
Gemini Deep Research — useful for external research that gets linked back into the vault

Step 1: Build Your Vault Structure

Structure determines how well AI can navigate your notes. Use numbered top-level folders so they sort predictably and communicate priority at a glance.

00_Home/          ← Dashboard, goals, quick capture
10_Periodic Notes/
  Daily/
  Weekly/
  Monthly/
  Quarterly/
  Yearly/
20_Learning/      ← Study notes, deep dives, technical concepts
40_Career/        ← Job search, behavioral prep, companies, strategy
50_Personal/      ← Self-reflection, life goals, personal projects
60_Reference/     ← Evergreen reference docs (setup guides, gear lists, etc.)

The numbering isn't decoration — it means your most important folders load first, and the AI navigating your vault hits the most personally meaningful content early.

A note on gaps: This vault skips 30. That's intentional — leave room to add folders without renumbering everything. Reserve number ranges for expected future categories.

Step 2: Build Your Daily Note Template

This is the foundation. Everything else depends on having rich daily notes that accumulate over time.

The most important principle: fill it out organically first. Then supplement with AI.

The daily note is valuable because it's yours — your actual words, your actual mood, your honest account of the day. AI input is most useful as a supplement to that raw material, not a replacement for it. A daily note filled entirely by AI tells the AI nothing about you.

A ready-to-use template is included in this repo at templates/daily-note.md.

Sections to include:

## ⏪ Yesterday's Recap
- Summary of the day prior
- What went well
- What could have gone better

## 🌅 Morning Check-in
- Mood and energy (numeric, so Dataview can track trends)
- How you feel physically and mentally — be specific
- Today's non-negotiables (2–3 things that *must* happen)
- Today's schedule (morning / afternoon / evening blocks)

## 💻 Domain-specific prep sections
(Whatever is relevant to your current phase — job search, study, creative work, etc.)

## 🔁 Habits
- Checkboxes for recurring behaviors you're tracking
- Include a text field for habits that benefit from reflection
  (e.g., "Threw something away — ___" rather than just a checkbox)

## 📝 Notes & Thoughts
- Unstructured brain dump — capture anything throughout the day
- Don't filter here. Let it be messy.

## 🌙 Evening Review
- What I accomplished today
- What I learned
- One win
- One thing I'd do differently
- Top priorities for tomorrow

## 🗂️ Notes & Activity (Dataview auto-populated)
- Notes created today
- Notes visited today

The evening review is the highest-leverage section. Even five minutes of honest reflection compounding daily produces more self-awareness than most people develop in years.

Step 3: Start Filling Notes — Organically, Every Day

Before involving AI at all, build a minimum of two to three weeks of consistent daily notes. This is the training data. The richer and more honest the notes, the more accurate the AI's understanding of you will be.

Things that make daily notes more valuable over time:

Be specific about feelings. "Felt tired" is useless. "Felt unfocused and irritable after a bad night of sleep, recovered after the gym" is useful.
Note patterns as you notice them. If you had a flow state, say what triggered it. If motivation tanked mid-afternoon, say why.
Capture small wins. The daily win field feels trivial at first. It compounds.
Include the uncomfortable parts. The notes that captured avoidance, frustration, and uncertainty are the ones that eventually led to the most useful AI analysis.

Step 4: Expand to Weekly, Monthly, Quarterly, and Yearly Notes

Once you have several weeks of daily notes, create periodic templates for longer time horizons. Ask the AI to generate retrospective notes for past periods by reading your existing daily notes.

Weekly notes should capture: theme of the week, key wins, main challenges, carry-forwards, and one honest line about trajectory.

Monthly notes zoom out further: what moved, what stalled, where energy went versus where it should have gone.

Quarterly and yearly notes are the altitude at which patterns become unmistakable. The same weakness that shows up in a Tuesday daily note shows up again in a quarterly review, and seeing that repetition is where real change starts.

Templates for weekly, monthly, quarterly, and yearly notes are included in the templates/ folder.

Step 5: Back Up Your Vault (Optional)

Non-technical users: You can skip this step entirely. Store your vault folder in iCloud, Dropbox, or Google Drive for basic backup. If you want sync across multiple devices without any setup, Obsidian Sync ($10/month) handles everything inside the app. Come back to this step later if you want full version history.

For technical users — Git backup:

Before your vault grows, back it up with Git. A vault with months of personal notes and no version control is a liability.

# In Terminal, from your vault root:
git init
git remote add origin git@github.com:yourusername/your-vault-repo.git
git add .
git commit -m "initial vault backup"
git push -u origin main

Then install the Obsidian Git plugin and configure it to auto-commit and push every 10–15 minutes. Key settings:

Auto save interval: 10 (minutes)
Auto push on commit: enabled
Pull before push: enabled
Commit message: "vault backup: {{date}}"
Disable popups for no changes: enabled

Add the included .gitignore to keep noise out of version history. It excludes Obsidian workspace state, plugin binaries, cache files, and other non-content files that pollute your commit history.

Your vault now auto-backs up to a private GitHub repo every 10 minutes. You can restore any version, sync to other devices, and never lose a note.

Step 6: Give the AI Access to Your Vault

This is the step that changes everything.

In Claude's desktop app with Cowork mode (or any AI with local folder access), select your vault's root folder as the working directory. The AI can now read every note in your vault — your daily reflections, your goals, your career notes, your study material, your personal history.

What this enables immediately:

Ask "what should I focus on today based on my recent notes?" and get an answer grounded in your actual life, not generic advice
Ask "what patterns do you notice in my daily notes over the past month?" and receive a genuine analysis
Ask the AI to create new notes, update existing ones, link related notes, or reorganize content

What it enables over time:

The AI develops an increasingly accurate model of how you think, what motivates you, and where your gaps are
Advice becomes personalized rather than generic
The AI surfaces connections you haven't made yourself — the same pattern showing up across your career notes, your daily reflections, and your habit tracking is something a human reader would miss

Important: The AI is only as good as what you put in. A vault full of AI-generated summaries tells the AI nothing about you. A vault full of your own words — imperfect, honest, consistent — is where the real value lives.

Step 7: Have the First Big Conversation

Once you have a meaningful vault and the AI has access, initiate a structured self-assessment. Don't ask vague questions like "what do you think of me?" — be specific and push back on the answers.

Good starting prompts:

"Given everything in my vault, how would you describe me? What are my genuine strengths and weaknesses?"
"What do you notice that I seem to be unaware of or unwilling to acknowledge?"
"What are my biggest opportunities given who I actually am — not who I'd like to be?"
"What concrete things should I do this week that would have the biggest impact on my short-term and long-term goals?"

Then challenge the responses. Ask "Is it possible you're misjudging me based on limited context?" Push for specifics. Ask it to separate what it learned from your organic notes versus what you built collaboratively with AI assistance.

Capture the output in a self-analysis note (e.g., 50_Personal/Self-Analysis - March 2026.md) and use source: collaborative in the frontmatter to flag it as AI-assisted.

Step 8: Build Domain-Specific Notes as Needed

The vault grows most naturally when you're actively working on something. Don't try to build everything at once.

Career phase: Create a career folder with behavioral interview notes, company research, and a career direction document. Work through behavioral questions one at a time — give the AI your real story, let it help you structure it, then refine the language.

Learning phase: When studying a new concept, create a note in 20_Learning/. Ask the AI to explain it deeply, include analogies and examples, add links to relevant resources, and connect it to prior notes. The result is a personal knowledge base, not just bookmarks.

Life audit phase: Use the AI to audit your hardware, software, habits, and routines. Ask "given everything you know about how I work, what am I underutilizing? What should I fix?" The recommendations will be grounded in your actual setup, not generic productivity advice.

Step 9: Track What's Yours vs. What's AI-Assisted

Add a source field to your note frontmatter:

---
source: organic          # Written entirely by you
source: collaborative    # Built together with AI
source: ai-generated     # Primarily AI output based on your input
---

This matters for two reasons. First, when you read a note back later, you know how much to trust it as a reflection of your own thinking. Second, it keeps you honest — if everything is ai-generated, the system is working against you rather than for you.

Principles That Make This Work

Consistency beats intensity. A two-minute evening review every day is worth more than a two-hour journaling session once a month. The AI needs time-series data, not one-time snapshots.

Be brutally honest in your notes. The most valuable analysis this system produces comes from notes that aren't flattering — notes about avoidance, about frustration, about patterns you didn't want to look at. Sanitized notes produce generic advice.

The vault is a mirror, not a manager. The system works best when you use it to see yourself more clearly — not to have AI tell you what to do. The AI surfaces patterns; you make the calls.

Organic input first, AI supplement second. Every section you fill out yourself before asking for AI input is worth ten AI-generated sections. Your own words are the raw material. AI helps you process and organize it.

Let it grow naturally. Don't try to build the perfect vault structure upfront. Start with daily notes and a few folders. The structure will reveal itself as you accumulate content.

The gap between knowing and doing is the real work. This system is exceptionally good at helping you understand yourself. What it can't do is close the gap between insight and action. That part is still yours.

What This System Has Produced (Real Examples)

To make this concrete — here's what this system generated in practice over roughly two days of active use with a mature vault:

Self-Analysis — a structured breakdown of strengths, weaknesses, biggest opportunities, and core misalignments, grounded in behavioral patterns from months of daily notes
Career Direction — a nuanced comparison of three career paths with compensation data, role fit analysis, and an honest risk assessment for each
Target Company Research — 10 companies ranked and analyzed against personal fit, compensation, work-life balance, on-call load, and remote viability
Behavioral Interview Prep — interview answers built from real career stories, improved iteratively through direct conversation
Technical Deep Dives — study notes on CS fundamentals and system design topics with real depth, linked to YouTube resources and prior notes
Hardware & Software Audit — a complete inventory with deployment status, optimization notes, and a prioritized recommendations tracker
Setup Recommendations Tracker — a living checklist of prioritized improvements, checked off and updated in real time

None of these would have been possible without the vault providing the context the AI needed to give non-generic, actually-useful output.

About

Built and documented by Aaron Chung — former SDE II at Amazon, currently building in public.

If this system works for you, open an issue or reach out. I'm iterating on it continuously.

This guide is itself a product of the system — written collaboratively, grounded in real experience.

Ai Coach