AWS · Cloud service
Amazon SageMaker with OrchestrAI
Catalog exported 2026-09-02
Run Amazon SageMaker from chat: training jobs, model registration, endpoint deployment, and inference.
OrchestrAI exposes 13 SageMaker operations: 8 are low-risk (read-only or low-impact), and 5 create or modify resources and run only after you confirm the plan. 4 of them also carry a step-level approval gate.
What teams use it for
ML teams use OrchestrAI to launch a training job, follow its status, register the resulting model, create an endpoint configuration, deploy an endpoint, send a test inference request, and list what is currently running. Training job and endpoint creation wait for your confirmation, deleting an endpoint or model is high risk with confirmation, and describe, list, and invoke are low risk. Updating an endpoint in place, processing jobs, notebooks, and deleting endpoint configurations are not covered.
Every SageMaker operation, with its risk level
| Operation | What it does | Risk | Step-level approval |
|---|---|---|---|
Create SageMaker Endpoint Config |
Create endpoint configuration | Low risk | No |
Create SageMaker Model |
Register a model | Low risk | No |
Describe SageMaker Endpoint |
Get endpoint details | Low risk | No |
Describe SageMaker Training Job |
Get training job details | Low risk | No |
Invoke SageMaker Endpoint |
Invoke endpoint for inference | Low risk | No |
List SageMaker Endpoints |
List endpoints | Low risk | No |
List SageMaker Models |
List registered models | Low risk | No |
List SageMaker Training Jobs |
List training jobs | Low risk | No |
Create SageMaker Endpoint |
Deploy an endpoint | Creates resources | Yes |
Create SageMaker Training Job |
Create a training job | Creates resources | Yes |
Stop SageMaker Training Job |
Stop a running training job | Creates resources | No |
Delete SageMaker Endpoint |
Delete an endpoint | Modifies existing | Yes |
Delete SageMaker Model |
Delete a model | Modifies existing | Yes |
Risk tiers come from the catalog: low is read-only or low-impact, medium creates resources and is reversible, high modifies existing resources, destructive may lose data. Every plan that creates or changes resources is shown with its cost estimate and waits for your confirmation. Operations marked with a step-level approval pause again on their own step. Destructive operations require a typed risk phrase.
Prompts that work
- Start a SageMaker training job named churn-xgb-v3 using the xgboost image with data from s3://ml-data/churn/train/
- Deploy the churn-xgb-v3 model to an endpoint called churn-prod on one ml.m5.large
- Invoke the churn-prod endpoint with this CSV row and show the prediction
Before anything runs
Every mutation shows its plan, cost estimate, and blast radius, then waits for your confirmation. Destructive operations require a typed risk phrase. Credentials are minted per run through OIDC federation and discarded afterward; nothing you create here is invisible later, because every resource lands in the desired-state ledger where drift is detected and can be converged. Details on the security page.
Frequently asked questions
- Can OrchestrAI update a SageMaker endpoint to a new model?
- Not in place; it can create a new endpoint config and endpoint, then delete the old endpoint after confirmation.
- Can OrchestrAI stop a running SageMaker training job?
- Yes, stopping a training job is medium risk and shows the job name before it proceeds.
- Which SageMaker operations need an extra approval step?
- 4 operations carry a step-level approval gate on top of plan confirmation: Create SageMaker Endpoint, Create SageMaker Training Job, Delete SageMaker Endpoint, Delete SageMaker Model. None of them is classed destructive.
Other AWS services
Related integrations
Try it on your own account
Connect your cloud read-only and see your resources, drift, and costs before anything runs. $5 minimum to start. Unused credits refunded in your first 14 days.
Unused credits refunded in your first 14 days.