Kit for AI

Description

Kit for AI is drop-in memory infrastructure and a context-engineering platform for AI agents and apps. It combines persistent memory, document ingestion, knowledge bases, semantic search, and connectors behind a single API and native MCP tools your agent calls directly, so there's no vector database, embedding pipeline, or chunking to build and babysit.

For memory, agents can remember, recall, and search facts as MCP tools mid-conversation, keeping users, preferences, and decisions across every session, with automatic deduplication and versioning. For knowledge, you can drop in any file or URL, PDF, Word, Excel, PowerPoint, CSV, HTML, images (via OCR), and YouTube videos, and get clean Markdown out, chunked, embedded, and searchable. URLs are stripped of nav, ads, and footers to return just the main content, and Pro plans can extract typed JSON to a custom schema. Scheduled URL refresh keeps sources current.

Knowledge bases group documents your AI answers from with cited sources, backed by hybrid vector-plus-keyword retrieval that reranks the best passage first. Because the agent pulls only the passages it needs instead of dumping whole documents, Kit for AI reports up to 90% fewer tokens on real workloads. The same tools work in Claude, Cursor, and any MCP agent with a one-command install, and over REST for any model.

Kit for AI is private by default: documents stay yours, are encrypted at rest, isolated per project via Spaces, never used as training data, and deletable for good. It offers a free Freemium tier, plus Pro and Business plans and custom Enterprise, with an official SDK and a live API playground.

Kit for AI's Core Features

• Persistent memory (remember, recall, search) over REST and MCP

• Document and URL conversion to clean Markdown or JSON

• OCR for scanned pages and images

• YouTube transcript ingestion for chat and search

• Knowledge bases with cited, grounded answers

• Hybrid semantic plus keyword search with reranking

• MCP-native tools for Claude, Cursor, and any agent

• Up to 90% fewer tokens via targeted retrieval

How to use Kit for AI?

1. 1

   Sign up: Create a free account and generate an API key, no credit card required.

2. 2

   Ingest: Drop in files, URLs, or YouTube links, or call the convert API.

3. 3

   Build a knowledge base: Group converted docs into a searchable knowledge base.

4. 4

   Connect your agent: Install the MCP server in Claude or Cursor, or use the REST API.

5. 5

   Use it: Have your agent remember facts and pull cited answers from your knowledge base.

Kit for AI's Use Cases

• Agent memory
• RAG knowledge bases
• Document conversion
• URL and web ingestion
• Structured data extraction

FAQ from Kit for AI

• What is Kit for AI?

  It is a memory and knowledge layer for AI agents that adds persistent memory, document ingestion and retrieval, knowledge bases, and connectors over one API and MCP, with no RAG stack to babysit.

• What file formats does it convert?

  PDF, Word, Excel, PowerPoint, CSV, HTML, images (via OCR), and plain text or Markdown. All convert to clean Markdown, and Pro can extract structured JSON to your own schema.

• Can it turn a URL into clean text?

  Yes. Paste any URL and it fetches the page, strips the nav, ads, and footers, and returns the main content as clean Markdown.

• How does the memory work?

  Agents call remember, recall, and search as MCP tools mid-conversation to keep users, preferences, and decisions across sessions, with near-duplicate dedup and versioning.

• Does it work with my model?

  Yes. The same tools work in Claude, Cursor, and any MCP agent with a one-command install, and over REST for any model.

• Is my data private?

  Yes. Documents stay yours, are encrypted at rest, isolated per project via Spaces, never used as training data, and deletable for good.