Skip to main content

How qbash Works

qbash is an orchestration layer for AI across your team. It connects multiple AI models, your data, your tools, and your people in one place - without locking you into any single model provider.

The core idea: instead of each person crafting their own prompts and picking their own models, your team builds shared Projects, Tasks, prompts, and knowledge bases that everyone can run consistently.

The building blocks​

Each piece below has its own section with the full detail — this is the one-line orientation and where to read more.

Models​

qbash connects to models from Anthropic, OpenAI, Google, xAI, and OpenRouter. You choose the model and switch at any time; your prompts, tasks, projects, and settings stay exactly as they are. See Enabled Models.

Prompts​

A prompt is a reusable, versioned AI instruction — write it once and publish, then anyone can run it from chat, call it from a Task, trigger it via API, or use it as a tool in a Project. See Prompts.

Chat​

Chat is the main interface most people work from: pick a model or a library item, type, and iterate in a conversational thread, with file attachments, tool use, and collaborative sharing. See Using Chat.

Projects​

A Project is a workspace that bundles instructions, prompts, Tasks, connectors, and memory folders with a default model, so everything is already active when you start a chat inside it. It is the highest-level way to organize AI work around a purpose. See Projects.

Tasks​

A Task is a Python automation that runs on qbash's managed runtime — calling prompts, using connectors, reading and writing memory, and returning structured output. Tasks run manually, on a schedule, via API, from triggers, or as tools inside Projects, with production logs, retries, and revision history. See Tasks.

Memories​

Memories are folders of documents your Projects, Tasks, prompts, and chats read and write as persistent, shared context. Make a folder Searchable by AI and it becomes a semantic knowledge base; it can also be exposed to external tools over MCP. See Memories.

Playgrounds​

Playgrounds run multiple prompt configurations side-by-side on identical inputs — across versions or across models — with the outputs in columns to compare. Save the run state and share the link to document a model or prompt decision. See Playgrounds.

Connectors​

Connectors link qbash to external services, usable as a Project tool, a Task integration, a prompt tool, or a trigger. See Connectors.

How they connect​

Models          → run inside Chat, Prompts, Projects, and Tasks
Projects        → workspaces that bundle instructions, tools, knowledge, and model settings
Tasks           → Python automations with triggers, logs, retries, and outputs
Prompts         → reusable instructions, run in Chat, Projects, or Tasks
Chat            → conversational interface for models, prompts, and project work
Memories        → persistent storage, searchable by Projects and Tasks
Playgrounds     → side-by-side testing of Prompts across models and versions
Connectors      → external service links, usable in Projects, Tasks, Prompts, and Chat

A typical team setup: admins connect the services the company uses, create Projects for major work areas, attach the right prompts, Tasks, connectors, and memory folders, and let team members work from those Projects without rebuilding context every time.

What qbash is not​

qbash is not a code assistant or a general-purpose chatbot. It is an operations layer - the infrastructure your team uses to build, share, govern, and run AI consistently.

Last updated on