Kit for AI è un'API per la memoria persistente degli agenti, la conversione di documenti e URL e il recupero con citazioni da basi di conoscenza. Supporta MCP e REST, con un piano gratuito e piani a pagamento su invito o personalizzati.
Kit for AI is an API and agent layer for persistent memory, document conversion, and grounded retrieval. It is designed to help AI agents keep track of facts across sessions while also turning files, URLs, and videos into searchable knowledge that the agent can use in conversation.
The product combines memory tools, conversion endpoints, and knowledge-base chat in one system. According to the site, agents can call its tools directly over MCP or REST, and the same API key works across both paths. That lets builders avoid wiring together a separate RAG stack, parser pipeline, and memory service when they only need a single retrieval and memory layer.
The documentation shows a focus on practical workflows: ingest content, convert it to Markdown or JSON, attach it to a knowledge base, and retrieve cited answers later. The pricing page also shows a free tier, invite-only Pro and Business plans, and custom Enterprise pricing for larger deployments.
Store facts, preferences, and decisions as durable memories that can be recalled later through MCP tools or REST endpoints. Near-duplicates are skipped server-side, and whole conversations can be processed to extract durable facts automatically.
Convert files or URLs into clean Markdown or structured JSON for LLMs and agents. The documentation lists support for PDF, Word, Excel, PowerPoint, CSV, HTML, images, and web pages.
Turn scanned pages and image inputs into text automatically through OCR, with optional deeper extraction for complex layouts on Pro plans.
Group converted content into named knowledge bases and query them with cited answers. The product also supports hybrid retrieval with vector and full-text search plus reranking.
Expose the same capabilities through MCP and REST so agents in tools such as Claude Code or other MCP clients can use the platform directly.
Process up to 25 files or URLs per batch request and attach results to a knowledge base in the same workflow.
Use remember and recall to keep user preferences, project facts, and decisions available across sessions, rather than re-asking for the same context in every conversation.
Feed files or URLs into the conversion API and attach the output to a knowledge base so an assistant can answer questions from grounded source material.
Convert invoices, forms, or other structured content into typed JSON using a schema when the goal is extraction instead of summarization.
Point an MCP-capable client such as Claude Code at the HTTP MCP endpoint so the agent can use memory, search, and conversion tools without a separate integration layer.
Use batch conversion and scheduled URL refresh to keep a library of reference material current for retrieval workflows and internal knowledge bases.
Kit for AI can be used through either MCP or REST with the same API key. The setup page shows an HTTP MCP endpoint and a REST base URL, and the documentation also mentions an official SDK.
The documentation says it supports documents such as PDF, Word, Excel, PowerPoint, CSV, HTML, and images, plus URLs and YouTube videos. OCR is applied automatically for scanned pages and image inputs.
It can return either clean Markdown or structured JSON. JSON output is driven by a user-supplied schema, which makes it suitable for tasks like invoice or form extraction.
The pricing page shows a free tier and invite-only paid tiers. Pro and Business plans are invite-only, while Enterprise uses custom pricing and includes options such as SLAs, dedicated infrastructure, SSO, custom integrations, and white-glove onboarding.
The setup page says it supports an optional end_user_id so each end user can have an isolated memory namespace, which is useful for multi-tenant apps. The account also has its own memory namespace if end_user_id is omitted.
I dati sul traffico sono solo a scopo di riferimento.
evomap.ai
EvoMap è un’infrastruttura per l’auto-evoluzione dell’IA che consente di condividere, convalidare ed ereditare le capacità degli agenti tra modelli e regioni. Offre un livello gratuito, i piani a pagamento Premium e Ultra e l’accesso tramite account a uno spazio di lavoro per l’evoluzione dell’IA.
venice.ai
Venice è una piattaforma di IA privata per chat, immagini, video, audio, codice, ricerca e flussi di lavoro con agenti. Offre un piano gratuito, piani a pagamento e un’API compatibile con OpenAI con diverse modalità di privacy.
arcade.dev
Arcade è un runtime MCP per agenti AI in produzione che combina autorizzazione degli agenti, strumenti affidabili e governance. Aiuta i team a collegare gli agenti ai sistemi reali, con opzioni di deployment flessibili e piani tariffari dal gratuito all’enterprise.
mulerun.com
MuleRun è una piattaforma di agenti AI per automatizzare i workflow end-to-end: ricerca, creazione di documenti e presentazioni, sviluppo di output web e monitoraggio proattivo. Disponibili piani gratuiti, a pagamento ed enterprise.
vectorize.io
Memoria open source per agenti, con contesto persistente e ricordi tra sessioni
leapility.com
Leapility aiuta creator, consulenti e specialisti a trasformare le proprie competenze in kit in abbonamento con basi di conoscenza, skill, playbook e agenti AI.
