Cuedra
A multi-tenant platform for building AI agents that answer from approved knowledge and call business APIs during a conversation.
What is Cuedra?
Cuedra is Imam Faheem's own multi-tenant AI agent platform. Businesses configure an agent with instructions, a model, approved knowledge, and API actions, then deploy it through an embeddable widget or API. Preview at cuedra.vercel.app.
Cuedra combines semantic retrieval over business content with live HTTPS API actions in the same conversation. An agent can answer a policy question from approved sources, then check an order or booking through the business's system. Public MCP servers are also supported as a preview feature.
The product includes a Node.js API and workers, a Next.js administration app, a Preact widget, and an Astro website. It is currently pre-launch while production onboarding and self-serve access are prepared.
Launch status
Pre-launch. The core agent, knowledge, action, publishing, widget, API, billing, and workspace systems are built; production onboarding and self-serve access are being prepared.
What stack does Cuedra use?
- RAG
- MCP
- Qdrant
- OpenAI
- LiteLLM
- Stripe
- Next.js
- Preact
- Node.js
- Express.js
- PostgreSQL
- Redis
- TanStack Query
- Zustand
- Webhooks
- API Actions
- SSE
- RBAC
What does Cuedra do?
01
Configure an agent with instructions, model choice, knowledge, and actions
02
Ingest documents, websites, and pasted content for semantic retrieval
03
Call HTTPS APIs with bearer tokens, API keys, or signed requests
04
Connect public MCP servers as a preview feature
05
Test draft agents privately and publish a version for visitors and API callers
06
Embed a Preact chat widget with one script tag
07
Call agents from another product through a REST API with scoped keys
08
Manage workspaces, roles, usage, and Stripe-backed plans
Screenshots
Next
Maktab