This repository contains examples of how to use Exabase with different use cases.
-
AI Flashcard generator: A simple example of an AI flashcard generator using Exabase, Next.js, AI SDK, and Tailwind CSS.
-
Sales Memory Copilot: A memory-first sales copilot using Exabase Bases and the Memory API (search + capture), with Next.js, AI SDK, and Tailwind CSS.
-
Personal assistant: A memory-first personal assistant demo with Exabase search + add memory, Next.js, tRPC, and AI SDK.
-
GitHub PR memory: Ingest public pull-request and review comments into Exabase memories, then chat with that code-review context; Next.js, tRPC, GitHub REST, AI SDK.
-
Simple Extraction Demo: Submit a URL or file to the Exabase Extraction API, poll for results, and browse extracted metadata, chunks, and attachments. Includes webhook delivery logs and reprocessing; Next.js, tRPC, shadcn/ui.
-
Document Diff Viewer Example: Submit two URLs or files concurrently, extract both, then compare side-by-side — metadata table with keyword diff and word-level text diff; Next.js, tRPC, shadcn/ui.
Memory – Give your agent a self-improving memory that gets smarter over time. Exabase handles storage, retrieval, and continuous refinement automatically.
Search – Deep multi-modal search across files, images, and documents. Highly configurable and batteries-included — no stitching together multiple tools.
Extraction – Turn PDFs, websites, and media into clean, standardised output your agent can actually work with.
Workers – Autonomous agents that continuously update your data, monitor the web, and keep your knowledge base current.
Bases – Instant filesystems with full version history. If your agent makes a mistake, roll back to any point in time.