Building a Personal Knowledge Agent
Turn your notes, documents, and bookmarks into something an agent can search and reason over — your own research assistant.
You've probably spent years collecting knowledge: meeting notes, project write-ups, saved articles, highlighted PDFs, hundreds of bookmarks you meant to read. Almost none of it gets used, because finding the right note at the right moment is harder than starting from scratch. The knowledge is there. Retrieval is the problem.
A personal knowledge agent fixes that. Point an agent at your collection with a few clear rules, and you can ask "what did I learn the last three times we ran a pricing change?" and get an answer built from your own notes, with links to the exact files. This lesson shows you how to choose a setup, get your material into a shape agents can use, write the instructions that make answers trustworthy, and keep the whole thing private and up to date.
What a personal knowledge agent actually is
A personal knowledge agent is three things together: an agent that can search and read, your collection of notes and documents, and a set of rules for how to use them. It does three jobs:
- Find: locate the note, document, or bookmark you half remember.
- Connect: spot patterns across material you collected at different times ("you've written about onboarding problems in 11 notes since 2023").
- Answer with citations: answer questions from your material, pointing to the exact files so you can check.
It's different from the memory features built into chat apps, which store a handful of facts about you (see memory for agents). A knowledge agent works over hundreds or thousands of your own files, and you control exactly what's in them. Think of it as the inward-facing twin of the research agents from the start of this module: those search the web, this one searches what you already know, and the same habit of tying every claim to a source applies.
Nadia, an independent management consultant, is our example. She has eight years of client notes, a folder of reports she wrote, around 2,000 browser bookmarks, and a habit of writing a short reflection after every project. Before every new engagement, she spends a day trying to remember what she's learned from similar ones.