Ground Truth and A2I
A private workforce for your Ground Truth and A2I workflows
SageMaker Ground Truth and Amazon A2I keep running. The Amazon Mechanical Turk workforce option does not. Add our trained team to your private workforce and keep your pipeline as it is.
What AWS changed
- ·Ground Truth and A2I are not closing.
- ·The Amazon Mechanical Turk workforce is no longer available in labeling jobs or human review workflows. It ended with MTurk on 2026-09-30.
- ·Jobs and workflows that used it need another workforce: a private workforce or a vendor workforce.
- ·Your task templates, S3 data and code stay. The workforce setting changes.
Source: AWS SageMaker documentation, checked 2026-09-30.
How a private workforce works
- ·A private workforce is a group of people you choose. You manage it with Amazon Cognito or your own identity provider (OIDC).
- ·You put people into work teams. Each labeling job or human review workflow sends tasks to one work team.
- ·People sign in to a labeling portal for your AWS account. They see only the tasks sent to their team.
- ·You control access. You can remove a person at any time and limit sign-in to set IP ranges.
- ·We staff it. You add our people to a work team. We train them on your task, and a team lead and QA reviewers run the day to day.
What we take on, and what goes elsewhere
- $5/hr Textract document review, text classification, AI output rating and correction, and custom text tasks.
- Custom rate Image, video, 3D point cloud and sensor labeling jobs, through the enterprise team. Talk to the enterprise team.
- We do not take graphic or violent content review, including moderation queues that contain it.
Move this week in five steps
- 1
Find what used the MTurk workforce.
List your labeling jobs and human review workflows (flow definitions). Look for the public workforce ARN, which ends in workteam/public-crowd/default. Search your infrastructure code for it too.
- 2
Set up your private workforce.
In the SageMaker console, open Labeling workforces, then Private. If you do not have one yet, create it with Cognito or your OIDC provider. Each AWS account has one private workforce per Region, and it can hold several work teams.
- 3
Create a work team for us.
Add the work emails we send you. Each person gets an invitation to your labeling portal. Add IP limits if your policy needs them.
- 4
Point your work at the new team.
For Ground Truth, start new labeling jobs with the private work team ARN. You no longer need the public workforce task price. For A2I, existing flow definitions cannot be edited. Create a new human review workflow with the private work team ARN and the same task template, then update your code to use the new workflow ARN.
- 5
Run a small batch first.
Send a test batch, check our answers against your gold set, then move the rest.
You can also staff your private workforce with your own employees, or subscribe to a vendor workforce in AWS Marketplace. Pick what fits your volume, data and budget.
Start a pilot →FAQ
Do we need to rebuild our pipeline?
No. You change the work team your jobs and workflows point to. For A2I, that means a new human review workflow. Your task template stays the same.
Does our data leave our AWS account?
Your data stays in your S3 buckets. Our people view tasks in your labeling portal, on accounts you control.
Can you handle our volume?
Tell us your weekly volume in the pilot form. We size the team in the pilot plan.
What does it cost?
$5 per hour per person for the task types above. Image, video, 3D and sensor jobs are priced by the enterprise team.