SPIRITTScale AI

Scale AI
Labels in, models out

SPIRITT creates Scale AI labeling tasks from your data, manages batches, ontologies and teams, and pulls finished labels into training the moment tasks complete.

SPIRITT Workspace input reading Connect me to Scale AI, with the Scale AI icon in a compact app tile.

What you can do with Scale AI.

  • Every data type covered

    Create image, segmentation, video, lidar, named entity, text collection and document transcription tasks.

  • Batches under control

    Group tasks into batches, set priorities, define ontologies and assign projects to your team.

  • Completion that wakes the pipeline

    Task callbacks reach SPIRITT when work finishes, so labels move into training without polling.

Get started in three steps.

01

Open your workspace

Start a SPIRITT workspace for your data operations, with your datasets, label specs and training targets.

A workspace connected to structured data
02

Connect Scale AI

Connect Scale AI with your API key so SPIRITT can create tasks, manage batches and read results.

SPIRITT Workspace input reading Connect me to Scale AI, with the Scale AI icon in a compact app tile.
03

Hand over labeling ops

Describe the data and the label spec. SPIRITT runs the pipeline from upload to training set.

Data, charts, and recurring workflow arrows

Training data on tap.

Questions

Scale AI, with SPIRITT.

01What can SPIRITT do with Scale AI?+
Create image, segmentation, video, lidar, named entity, text collection and document transcription tasks, upload or import files, manage batches, priorities, ontologies and project defaults, tag tasks and manage team assignments.
02Can finished tasks trigger the next step?+
Yes. Tasks carry a callback address, so Scale notifies SPIRITT on completion and SPIRITT moves the labels into training immediately. Missed callbacks can be resent.
03How does it track progress?+
SPIRITT reads batch status and completion counts and posts progress and finish estimates on your schedule.
04Can it watch quality?+
Yes. SPIRITT reviews audits and labeler training results and flags projects where quality slips.
05Where do labels go?+
Into BigQuery tables, Hugging Face datasets, Supabase or your training pipeline, with metadata linking each label to its source.
06How do I get started?+
Open a SPIRITT workspace, connect Scale AI and share your data location and label spec. SPIRITT starts with a pilot batch.
Buy from builders who use what they sellBuilt usingSPIRITT