proposed stage so a human can review the result before it is finalized. This model lets you automate high-frequency, low-risk operations while keeping a human in the loop for anything that changes sensitive data.
Retrieve your workspace
Fetch all teams and tasks that belong to your tenant and organization in a single call.array
List of Human-AI teams scoped to your tenant and organization.
array
The 100 most recent AI tasks across all teams, ordered by creation date descending.
Create a Human-AI team
Teams are the containers for collaborative work. Only users with the Admin or Manager role can create them.Team request fields
string
required
A short, human-readable label for the team. Must be non-empty.
string
An optional longer explanation of what this team handles.
integer
How often (in minutes) the team’s AI agent runs its work cycle. Defaults to
30 if omitted or set to 0.object
A JSON object that defines who must approve proposed outputs from tasks in this team. The structure is flexible — for example,
{ "required_role": "MANAGER" } requires a Manager-level user to sign off. Omit or pass {} for no policy enforcement at the team level.Team response fields
string
UUID that uniquely identifies the team.
string
The team’s display name.
string
Optional description of the team’s purpose.
string
Lifecycle status of the team. Starts as
active.integer
How often the AI agent runs, in minutes.
object
The approval rules attached to the team.
string
ISO 8601 timestamp when the team was created.
string
ISO 8601 timestamp of the last update.
Create an AI task
Tasks are units of work you assign to a team’s AI agent. Users with the Admin, Manager, or Editor role can create tasks.Task request fields
string
required
UUID of the team this task belongs to. Must reference a team within your tenant and organization.
string
required
A concise description of what the task involves. Must be non-empty.
string
Task urgency level. Accepted values:
LOW, NORMAL, HIGH, CRITICAL. Defaults to NORMAL.object
A free-form JSON object that provides the AI agent with the data it needs to complete the task — for example, a SKU, a date range, or a customer ID. Defaults to
{}.boolean
When
true, the task pauses at proposed status after the AI generates output, waiting for a human to approve or reject it before the result is applied.string
Optional deadline for the task, in RFC 3339 format (e.g.
2024-11-20T17:00:00Z).Task response fields
string
UUID that uniquely identifies the task.
string
UUID of the team the task belongs to.
string
The task description.
string
Current lifecycle status (see below).
string
Task urgency:
LOW, NORMAL, HIGH, or CRITICAL.object
The data payload supplied when creating the task.
object
The AI-generated result, populated once the agent completes its work.
boolean
Whether human sign-off is required before the output is applied.
string
Optional deadline in RFC 3339 format.
string
Timestamp when the AI agent began processing, if started.
string
Timestamp when the task reached a terminal state, if completed.
string
ISO 8601 timestamp when the task was created.
string
ISO 8601 timestamp of the last update to the task.
Task lifecycle
Every task moves through a defined set of statuses from creation to completion.1
pending
The task has been created and is waiting for the AI agent to pick it up during the team’s next cadence cycle.
2
in_progress
The AI agent has started working on the task. The
started_at timestamp is set at this point.3
proposed
The AI agent has finished and written its result to
proposed_output. If requires_approval is true, the task waits here for a human reviewer to approve or reject the output before it takes effect.4
completed
The task is done. If approval was required, it was granted. The
completed_at timestamp is set.A task with
requires_approval: false skips the proposed stage and moves directly from in_progress to completed once the agent finishes.Approval policy reference
Theapproval_policy field on a team is a flexible JSON object. You define the structure that matches your internal workflow. A common pattern is role-based approval:
approval_policy key conventions
approval_policy key conventions
These keys are illustrative — Synq stores the object as-is and your approval workflow layer enforces the policy.