Research & Extraction
Search thousands of files in seconds. The agent uses Grep, Glob, and File Read to find exactly what you need, extract the relevant pieces, and compile a structured summary.
Impact
What changes
Without Lapu AI
A project manager spends an hour opening 30 meeting note files one by one, searching for budget discussions, copying relevant paragraphs, and assembling a summary in a new document.
With Lapu AI
Lapu AI greps all 30 files for budget-related content in seconds, reads the surrounding context, and produces a structured summary with source references — ready in under 5 minutes.
The challenge
Important information is scattered across hundreds of files — meeting notes, project docs, config files, log archives, email exports, legal documents. Finding a specific detail means opening files one by one, searching manually, and piecing together context from multiple sources. System search tools find filenames but cannot understand content or compile findings.
How Lapu AI solves this
Lapu AI uses Glob to locate files by pattern, Grep to search contents with regex, and File Read to pull relevant sections — across your entire local file system. It can search thousands of files in seconds, extract matching passages, cross-reference findings across documents, and compile everything into a structured summary. The agent understands what it reads, so it can answer follow-up questions, filter by relevance, and highlight connections you might miss.
Searches run locally on your machine. Only the context needed for AI reasoning is sent to model providers.
Workflow
How it works
Define your search scope
Tell the agent where to look and what to find — for example, 'Search all markdown files in ~/projects/ for mentions of authentication and summarize what each project does for auth.' The agent uses Glob to locate matching files.
The agent searches and reads
Using Grep for content matching and File Read for context, the agent scans the target files, identifies relevant passages, and reads surrounding context to understand the full picture.
Get a compiled, structured output
The agent produces a summary — organized by source, with citations, key findings, and extracted data. It can write the results to a file, answer follow-up questions, or refine the search based on what it found.
Try it yourself
What you would type
Copy any of these into Lapu AI to get started immediately.
>Search all meeting notes in ~/notes/ for any mention of the Q2 budget and compile the relevant decisions into a single summary.
>Find every config file across my projects that references a database connection string and list them with the connection details.
>Search the legal folder for all clauses related to termination across our client contracts and compare them side by side.
Ready to try this workflow?
Download Lapu AI and run it on your own machine. Free to start.
Download for freeFAQ
Common questions
How fast is it compared to system search?
The Grep and Glob tools use ripgrep under the hood, which is extremely fast — it can search tens of thousands of files in seconds. Unlike system search, the agent also reads and understands the content, so it can summarize and cross-reference findings rather than just listing file matches.
Can it search inside PDFs and Office documents?
For text-based files (markdown, txt, csv, json, code), Grep searches directly. For PDFs and Office documents, the agent first uses built-in document skills to extract text, then searches the extracted content. It handles the format conversion automatically.
Does it work across different file types at once?
Yes. A single search can span any combination of file types — markdown notes, JSON configs, Python scripts, CSV data, text logs. The agent adapts its reading strategy based on what it finds.
Explore more
Related use cases
File Organization
500 files in your Downloads folder? The agent reads them, understands what they are, sorts them into labeled folders, and renames everything — in minutes, not hours.
See how it worksDataData Processing
Describe the transformation you need in plain English. The agent writes and runs the Python script, shell pipeline, or Node job — you just approve and get clean output.
See how it works