Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.quiva.ai/llms.txt

Use this file to discover all available pages before exploring further.

Overview

QuivaWorks is built around a small set of core concepts that work together to create an intelligent AI platform for teams. Understanding these will help you build effective assistants, design automations, and get the most out of the platform.

Assistants

Assistants are the heart of QuivaWorks — AI collaborators configured for specific tasks, roles, or domains. Unlike general-purpose AI, QuivaWorks assistants retain context, access your systems, and improve through intentional refinement.

Team vs. Personal Assistants

Team Assistants

Accessible to everyone in your account. Use for standardised workflows, shared tools, and company-wide resources.

Personal Assistants

Private to the creator only. Use for individual work, experiments, or personal use cases.

Configuration Layers

Every assistant has two configuration layers that merge at runtime:
  • Team Settings — Shared instructions, knowledge, integrations, and context variables for the entire team
  • Personal Settings — Individual configurations that layer on top of team settings without affecting others
Account-Level Settings sit above both and apply to all your assistants when enable in each assistants settings:
SettingPurpose
Global InstructionsDefault instructions applied to all assistant invocations
Global KnowledgeCompany-wide knowledge sources, enabled per assistant
BrandingCustomise the look and feel of the interface

Assistant Configuration

Each assistant is configured across four areas:

Instructions

Define the assistant’s role, personality, tone, and responsibilities

Knowledge

Give the assistant access to your documentation and context

Integrations

Connect to your systems via MCP protocol

Context Variables

Configure environment-specific parameters

Learn more about Assistants

Dive deeper into creating and configuring assistants →

Flows

Flows are automated sequences of steps that execute in response to a trigger. Think of a flow as a recipe for automation — triggers start the process, steps perform actions, and variable mappings pass data between steps.

Flow Structure

Triggers

Start flows via webhooks, schedules, HTTP requests, emails, file uploads, or stream events

Steps

Run assistants, make decisions, transform data, call APIs, loop over items, or wait for human input

Functions

Utility operations for key-value storage, object storage, streams, and data manipulation
Flows support conditional branching, parallel execution, nested flows, and human-in-the-loop approval gates.

Learn more about Flows

Explore triggers, steps, and flow configuration →

Integrations & MCP

Model Context Protocol (MCP) is an open standard that connects assistants and flows to external systems, tools, and data sources.

What MCP Enables

  • Pre-built integrations — Connect to popular business tools out of the box
  • Auto-generation Custom integrations — Create MCP servers automatically from OpenAPI specifications to build your own for proprietary systems and internal APIs
  • Enhanced specs — OpenAPI specs with embedded agent instructions are supported

Built-in Tools

Every assistant comes with a set of built-in tools available out of the box:
  • Web Search — Recent news and general web results
  • Content Fetching — Retrieve and parse URLs as markdown
  • Document Analysis — Search and analyse knowledge base documents
  • Math Evaluation — Solve mathematical expressions
  • JSON Extraction — Extract values using JSONPath
  • Regular Expressions — Pattern matching using Go RE2 syntax
  • Date/Time Parsing — Parse and format with timezone support
  • Data Encoding/Decoding — Convert data using various schemes
  • Cryptographic Hashing — Industry-standard hash generation
  • Task Management — Create and manage structured task lists
  • Escalation — Route requests to human supervisors when needed

Sub-Agents

For tool-heavy workflows, QuivaWorks automatically uses sub-agents to prevent context window overload. The context window is split across multiple tool calls, each handled by a dedicated sub-agent, with results consolidated back to the main assistant.

Learning System

Assistants improve through intentional refinement — not automatic adjustment. You stay in control of when and how your assistants evolve.
1

Vote on Interactions

Use thumbs up/down to flag what works and what doesn’t in real conversations
2

Review Insights

The Learning tab consolidates feedback into actionable patterns from successful interactions
3

Refine Behaviour

Update instructions, knowledge, or settings based on what you’ve learned
QuivaWorks uses intentional refinement by design. Assistants don’t change automatically — you decide when and how to update them based on the insights surfaced.

Learn more about Learning

See how to improve assistants through feedback and insights →

Collaboration

QuivaWorks is built for teams. Any assistant session can be shared with teammates in real-time — no separate tools or context-switching required.
  • @Mentions — Type @ in any session to invite team members instantly
  • Shared With Me — A dedicated sidebar section showing all sessions shared with you
  • Unread Badges — Track new messages across all shared sessions
  • Smart Notifications — Get alerts when teammates complete tasks or reply
  • Multi-Tab Conversations — Run multiple assistant sessions simultaneously; assistants continue working while you switch tabs

Learn more about Collaboration

See how to share sessions and work with your team →

How It All Works Together

1

Create an Assistant

Configure instructions, knowledge, and integrations for your use case. Set team settings for shared standards and personal settings for individual customisation.
2

Work Together

Use the assistant in sessions. Share with teammates using @, track updates with unread badges, and collaborate in real-time.
3

Automate with Flows

Build flows to trigger assistants automatically — on a schedule, via webhook, from an email, or any other trigger.
4

Improve Over Time

Vote on interactions, review the Learning tab, and intentionally refine your assistants based on the insights surfaced.

Next Steps

Create Your First Assistant

Build an assistant tailored to your use case

Explore Flows

Automate tasks with triggers and steps

Team Collaboration

Share sessions and work together in real-time

API Reference

Integrate QuivaWorks programmatically