Create a Cohere embed job
Approval-gated Cohere workflow for creating one async embed job from an existing dataset. It reads the input dataset and recent embed jobs first, then calls recued-core.cohere.embed_job.create with an explicit model, dataset ID, input type, optional name, embedding types, and truncate policy. It does not upload or delete datasets, cancel unrelated jobs, call streaming chat/generate endpoints, use connectors, mutate fine-tuned models, or call arbitrary Cohere routes.
How it works
Inspect the data fetches, transforms, gates, and output this recipe runs.
Process (11 steps)
normalized_dataset_id
trim
Trim whitespace from setting dataset id
normalized_job_name
trim
Trim whitespace from setting job name
normalized_reason
trim
Trim whitespace from setting reason
dataset
?
jobs_before_raw
?
create_job
?
jobs_before
default
Apply default
jobs_before_count
count
Count items in jobs before
job_rows
slice
Take a subset of
job_table
to_table
Format results as a data table
card
to_summary
Format results as a summary card
Settings
Configurable at install. Defaults shown — change them anytime in Recued.
cohere
setting
=
reason
setting
=
job name
setting
=
truncate
setting
=
END
dataset id
setting
=
input type
setting
=
search_document
embed model
setting
=
embed-v4.0
embedding types
setting
=
float
Trust & control
What installing this recipe would let it do. Recued grants these permissions at install — review them there before approving.