Nora Studio — Application Description
Purpose
Nora Studio is a team-oriented platform for building and running AI agents and workflows. It lets you manage knowledge bases (RAG), prompts, chat, scheduled jobs, and custom tools in one place—per project and per organization.
What It Does
-
Workflows
Create and run workflows that use AI agents. Workflows are defined at the organization level; you can create, open, and execute them from the Workflows UI. -
Chat
Per-project chat backed by a stream API (thread, system prompt, model settings). Supports project settings and API key–based access. Intended for conversation UIs and integration with your agents. -
RAG (Retrieval-Augmented Generation)
- Collections: Vector collections (Qdrant-backed "Higher RAG") for documents and DB-sourced data.
- Documents: Upload files → chunk (e.g. semantic strategy, size/overlap) → optional highlight → embed into a collection. Supports OCR and vision options.
- Database source: Connect a DB, pick tables/columns, apply filters, then embed table data into a RAG collection.
- Vectors: Inspect and manage vector indices and embeddings.
-
Prompts
Manage prompt collections and individual prompts per project. Run prompts with selected models and optional attachments (e.g. for testing and iteration). -
Memory
Per-project memory configuration: short-term (e.g. Redis-backed token/summary settings) and long-term (RAG embeddings withcontent_type: user_info). Supports profile-style cards and summary rates for chat/agent context. -
Tools
Define and manage tools (e.g. DB-backed, with parameters) for use by agents or workflows. -
Automation (Scheduling)
Create schedules that call a destination URL on a cadence (interval, period, run time, timezone, optional start/end). Configure method, body, headers, retries. Run on demand and view run logs. -
Organization & projects
Organizations contain projects. Each project has its own Chat, Logs, Prompt, Memory, RAG, Tools, Automation, and Members. Organization-level settings include general info, members, and usage. -
Access
Entry is email-based (landing → verification code → verify). Project and org membership control access to Chat, RAG, workflows, and the rest.
Technical Context
- Frontend: Next.js (App Router), React, Tailwind, HeroUI.
- Backend: Django; APIs for user, organization, project, RAG (Higher RAG), scheduling, prompt, agent.
- RAG stack: Qdrant (via nora_qdrant), embeddings and vector search, optional QA and tuning flows.
- Infra: Redis (cache/session/rate limit), optional Docker/Nginx, deployable (e.g. Railway).
In Short
Nora Studio is an application for creating AI agents and workflows and collaborating with your team: you manage RAG knowledge (documents and DBs), prompts, chat, memory, tools, and scheduled automation per project, and run agent-based workflows at the organization level.