Hi, I’m Kane, the developer behind Linkly AI. In this post, I’ll walk through 12 real scenarios where I use Linkly AI in my daily work — to show what’s actually possible when AI can access a library of 10,000+ local documents.
What My Document Library Looks Like
As a developer, product manager, and former engineering director, my work computer has accumulated over a decade of files: project documents, meeting notes, research papers, technical docs, work journals, personal reflections, HR files, corporate paperwork, and diaries.
I’m fortunate that I started using a Markdown editor (MWeb) over a decade ago for local writing and note-taking. That alone accounts for about half of my library — roughly 5,000 Markdown files.
I also use online tools like Notion, flomo, and Lark Docs. To keep everything searchable in one place, I periodically export content from these tools as Markdown or HTML files and drop them into Linkly AI.
My software project documentation is also written in Markdown, stored in directories alongside the code — so those naturally get indexed by Linkly AI too.
On top of that, I’ve collected a fair number of ebooks, PDF papers, and technical references. They’re fewer in number than my notes, but each one is far more substantial in content.
Since I started using Linkly AI, I’ve mostly stopped using online clipping tools. I installed a Chrome extension that converts any open webpage to a Markdown file with one click. Interesting pages get saved to a folder, automatically indexed by Linkly AI, and instantly searchable.
Over the years, all of this has added up to over 10,000 documents.
I’m honestly not a heavy knowledge-management user. Compared to lawyers, consultants, or academics, my library is probably modest, and I don’t use it every single day. I reach for Linkly AI most when I’m writing, doing retrospectives, or having in-depth discussions with AI about a specific topic.
Here are the 12 scenarios where I use Linkly AI with Claude the most:
Building a Targeted Resume
I’ve been attending more events and meetings recently, and sometimes I need to prepare a tailored resume on short notice. All my past resumes, promotion records, and project histories are already in my document library.
So I just tell Claude: “Build me a resume focused on my management experience.” It searches my documents, reads and filters the relevant ones, and drafts a new resume for me.
I particularly like this scenario:
- It’s a great test case — it requires searching, reading, comprehending, and summarizing across multiple documents, and since it’s about me, I can instantly tell if anything is wrong.
- AI tools don’t know who you are by default. I occasionally have AI build me a resume so it can better understand my background for future conversations.
- There’s also something surprisingly moving about it. Information about yourself triggers emotional resonance — it often surfaces memories from years ago that I’d completely forgotten.
Startup Retrospectives
Every startup project I’ve worked on has left behind documents — ideas, plans, summaries, retrospectives. When I want to look back on a past venture or distill lessons from a specific project, I ask Claude to search my library and compile a retrospective.
Here’s what makes this interesting: I like brainstorming startup ideas with Claude on claude.ai, and I tend to get carried away with unrealistic ideas. But because Claude has already read my past project documents and reflections, it often gives me sobering, perceptive feedback. Something like:
This idea sounds creative, but based on your reflections from the xx project, I’m not worried about your execution on the tech and product side — I’m more concerned that you might face similar go-to-market challenges…
It’s a remarkable feeling — you’re discussing the future with AI, and it responds with your own past. Honestly, it’s more grounded than advice from most friends, because it has actually read what you wrote at the time.
Finding Contracts, Corporate Documents, and API Keys
Over the past few years of building startups, I’ve registered companies in both China and the US, filed trademark applications, and accumulated various platform certificates and API keys for app store submissions.
These files sit untouched most of the time. But when you actually need them — for due diligence, tax filings, or updating business registration — you’re suddenly scrambling to find them.
I used to open Finder and rely on memory to navigate to the right folders. Now I tell Claude “find all PDF files related to trademark registration,” and it pulls up trademark acceptance notices, business licenses, and spousal consent letters — complete with application numbers and dates.
This is a very practical scenario: these files are needed maybe once or twice a year, but not finding them when you need them is genuinely frustrating.
Visa Documents
Last year when applying for a visa, I prepared a stack of materials: DS-160 confirmation, employment verification, bank statement checklists, interview preparation documents. After the application was done, these files sank into some folder and were never opened again.
This year, when I needed to prepare again, I asked Claude to find all my visa-related files. It didn’t just locate the documents — it read through my checklist and told me which materials I’d prepared last time and which ones needed updating.
My takeaway from this: we repeat certain tasks multiple times in life (visas, taxes, moving). Starting from scratch each time is wasteful. If AI can pull up your previous preparation materials for reference, the efficiency gain is significant.
Research for Writing
Writing is one of the highest-frequency use cases for Linkly AI — whether it’s blog posts, work summaries, legal analysis, academic papers, or industry reports, writing always requires extensive research and citation.
I’d previously collected information on over 600 AI tools, plus extensive notes and conversation logs about ChatGPT and AIGC. When I wanted to write about the AI tool ecosystem, going through these materials one by one was impractical.
So I had Claude use the explore feature to get a bird’s-eye view of my dedicated AI knowledge library, see what categories existed, then drill into specific directions (like “image generation AI tools”) to search, read, and extract key information.
The hardest part of writing isn’t having no material — it’s forgetting what material you’ve already collected. Linkly AI helps enormously here: it finds your own materials, summarizes them, and provides source links for proper citation.
Using a Dozen Ebooks as Startup References
My library includes several ebooks and lengthy PDFs about entrepreneurship and business management — like a 240,000-word CEO Financial Analysis and Decision-Making textbook. These books used to just gather dust because reading them end-to-end takes too long, and finding a specific concept means you have no idea which page to look at.
Now I can ask Claude to search within these books for “cash flow” or “balance sheet.” It tells me which chapters contain these concepts, on which pages, and shows me the surrounding context.
This turns multiple lengthy PDFs into a searchable knowledge base, rather than files you can only read sequentially.
Digging Up Old Weekly Reports
I worked at my previous company for over a decade and wrote hundreds of weekly reports and quarterly summaries. Once written, they were never opened again.
Until one day I wondered what projects I’d actually worked on in 2018 — and realized those weekly reports were the best timeline I had. I asked Claude to search for “my 2018 summaries,” and it found weekly reports from 2014 to 2019, quarterly reviews, and even presentation slides I’d completely forgotten about.
Seeing my own 2014 weekly report (“automated allocation plan, inventory API integration”) felt like archaeology. These aren’t just work records — they’re cross-sections of professional growth.
Helping a Friend Prepare for an Interview
A friend was interviewing for a product manager role and asked if I had any tips. In my previous role as a manager, I’d interviewed many candidates and had organized interview training sessions — all that material was on my computer.
I asked Claude to find everything related to “interviews” in my documents. It turned up interview records, interviewer training slides, candidate evaluation sheets, and my own interview strategy notes. Then I had it extract key takeaways and compile them into a reference document for my friend.
The core insight: your expertise is buried in documents you wrote but will never proactively open again. AI reactivates them.
Summarizing Management Experience
Similarly, when I wanted to systematically review what management methodology I’d accumulated, the experience was scattered everywhere: OKR training slides, promotion defense materials, team-building docs, quarterly reviews.
This kind of cross-document information synthesis is something humans are terrible at. You might vaguely remember writing about something in some document, but synthesizing a dozen documents into a coherent picture is practically impossible for the human brain.
Claude searches for “team management,” “OKR,” “performance,” “training” and other keywords, finds relevant documents, reads each one, extracts key points, and assembles a structured summary. It typically reads 10–15 documents for this.
It even summarizes my management style, philosophy, and blind spots — all derived from being able to access and understand my historical management documents.
Rediscovering Old Startup Ideas
As an entrepreneur, product ideas and business model concepts pop into my head constantly. Most get jotted down in Notion or discussed in ChatGPT conversations. Over time, these fragments get buried.
When I asked Claude to search for “startup ideas,” “product concepts,” and “business model,” it dug up a creative idea list from my ChatGPT conversation history that I’d completely forgotten: interactive experience apps, translation tools, workflow automation… Some still look interesting; others, in hindsight, were clearly impractical.
But the point isn’t that every idea is good — it’s that none of them are lost. When the right timing or context appears, they can be retrieved.
Reviewing What I’ve Been Working On
This is a small but practical scenario. Sometimes on a Friday afternoon, I want to review what I did this week, but I’m too lazy to browse through folders one by one.
Linkly AI recently added an explore feature that provides a bird’s-eye overview of your entire document library, including a “Recent Activity” section showing which directories had changes in the last 7 days. For instance, it tells me “linkly-ai-v3 directory had 4 file changes” — and I immediately recall what I focused on this week.
It’s not flashy, but it solves a real problem: which documents did I modify recently, and what recent context do they contain.
Tracing My Technical Learning Journey
This one is a bit personal. One evening I got curious: what technologies have I actually learned over all these years? So I asked Claude to search my MWeb notes and Hugo blog posts for technical content.
It found my 2017 tech sharing schedule (with React Native, Vert.x, and other topics that were hot at the time), plus technical learning notes I’d long forgotten.
This “archaeological” review may not have practical utility, but it reveals how your tech stack evolved — from PHP/Python to product management, to frontend, to Rust — with each transition driven by its own context and circumstances.
Some Reflections
After using Linkly AI for a while, I’ve come to believe its greatest value isn’t “search” itself — it’s that it makes your archived documents valuable again.
We all constantly produce documents — notes, emails, reports, contracts, retrospectives, musings — but once written, 99% will never be opened again. They sit in some folder, effectively nonexistent.
Every knowledge management theory, framework, methodology, and tool that came before was trying to solve: how to record? How to organize? How to find things later? But none truly solved the problem of how to use that knowledge. Not until large language models appeared — and with them, AI knowledge base products.
What Linkly AI does is actually simple: add your folders, and those sleeping documents naturally become searchable and readable by AI tools.
If you’ve also been accumulating documents on your computer for years, give it a try.
Linkly AI is a local document search engine supporting PDF, Markdown, Word, HTML, images, and more. All data stays on your computer — nothing is uploaded to the cloud. Through the MCP protocol, AI tools like Claude, ChatGPT, and Cursor can directly search and read your local documents.
Learn more: linkly.ai