Abbie
Abbie is the AI coworker and the first thing in the sidebar. You give her a task in conversation and she works through it, using the tools, knowledge and integrations available to your account. Abbie has five areas:
Abbie’s settings cover Rapport (tone and how she works with you), Profile, Skills, Integrations, Model pool, Synapse Memory, Environment, Outreach and Task filing.
Pipeline traces and diagnostics are kept for 7 days, independently of the conversation retention period set for the account.
Spaces
A Space is where a body of work lives. Open Spaces in the sidebar to create and browse them. Each Space has four sections:Files
Documents and attachments belonging to the Space
Tasks
Work items, with their own prefix so they read as
ABC-1, ABC-2, …Records
Structured data captured through configured record types and form views
Chats
Conversations scoped to this Space
Spaces overview
How a Space is put together and how to configure one →
Meetings
Meetings is its own top-level area. Meetings can be recorded and transcribed, and each plan includes a monthly allowance of recording and transcript hours — see Plans & Pricing.Assistants
Assistants are AI collaborators configured for a specific task, role or domain. Where Abbie is your general coworker, an assistant is shaped for one job and can be shared with the whole team.Team vs. personal assistants
Team assistants
Available to everyone in the account. Use for standardised workflows, shared tools and company-wide resources.
Personal assistants
Private to the creator. Use for individual work and experiments.
Configuration layers
Every assistant has two configuration layers that merge at runtime:- Team Settings — shared instructions, knowledge, integrations and context variables
- Personal Settings — your own configuration, layered on top without affecting anyone else
The four configuration areas
Instructions
Role, personality, tone and responsibilities
Knowledge
Documentation and context the assistant can draw on
Integrations
Your systems, connected over MCP
Context Variables
Fixed values the assistant should always know — a project name, a default region. Never credentials.
Learn more about assistants
Creating, configuring and refining assistants →
Flows
Flows are automated sequences that run in response to a trigger. A trigger starts the flow, steps do the work, and variable mappings pass data between them.Triggers
Webhooks, schedules, HTTP requests, emails, file uploads and stream events
Steps
Run assistants, branch on conditions, transform data, call APIs, loop, or wait for a person
Functions
Key-value and object storage, streams, and data manipulation
Learn more about flows
Triggers, steps and flow configuration →
Integrations and MCP
Model Context Protocol (MCP) is an open standard that connects assistants and flows to external systems, tools and data sources.- Pre-built integrations — connect to popular tools out of the box
- Generated integrations — build an MCP server from an OpenAPI specification for your own internal APIs
- Enhanced specs — OpenAPI specs with embedded instructions are supported
Built-in tools
These tools are available without any integration set-up:Research
Research
- Web search — general web results
- News search — recent news results
- Web fetch — retrieve and parse a URL
Data and computation
Data and computation
- Calculator — evaluate mathematical expressions
- JSON query — extract values using JSONPath
- Regex — pattern matching
- Date/time — parse and format dates and times
- Encoding — convert data between encodings
- Hash — generate hashes
People in the loop
People in the loop
- Escalate — route a request to a person when judgement is needed
Tools and connectors
Connecting assistants to your systems →
Sub-agents
For tool-heavy work, QuivaWorks splits the job across sub-agents so a single context window doesn’t have to hold everything. Each sub-agent handles part of the work and results are consolidated back.Learning
Assistants improve through intentional refinement — you decide when and how they change.1
Vote on interactions
Thumbs up or down on real conversations to mark what worked
2
Review insights
The Learning tab consolidates that feedback into patterns
3
Refine behaviour
Update instructions, knowledge or settings based on what you’ve learned
Assistants don’t change on their own. Nothing is applied until you apply it.
Learn more about learning
Improving assistants through feedback →
The account
The account is the tenancy boundary. Users, Spaces, assistants, flows, credits and storage all belong to one account, and data is isolated per account.- Users and roles — seven roles: root, admin, developer, monitor, billing, collaborator and client. Only root and admin can invite people or change roles.
- Account settings — details, branding, global instructions and knowledge, escalation responders, data retention and specialization. Open Account in the sidebar (root and admin only).
- Billing & Plans — plan, credit balance, usage log and top-ups.
Roles and permissions
What each role can do
Account settings
Everything that applies to the whole account
Collaboration
Abbie conversations can be shared with named people as Editor or Read-only, or with everyone in the account. The owner controls access and can change or revoke it at any time.Learn more about collaboration
Sharing conversations, access levels and @mentions →
Next steps
Getting started
Create your account and secure it
Create your first assistant
Build an assistant for your use case
Explore Spaces
Files, tasks, records and chats
Plans & pricing
Credits, storage and retention by plan