CoCo in pipelines

Introduction to Snowflake CoCo CLI

James Cha-Earley

Senior Developer Advocate, Snowflake

From answers to pipelines

An answer from CoCo, a snapshot taken once

From answers to pipelines

A script, which repeats the work when someone runs it

From answers to pipelines

A pipeline, which runs on its own inside Snowflake

Dynamic tables

The source table, SHOP.REVIEWS

Dynamic tables

A query on top that uses AI_CLASSIFY to give each review a topic

Dynamic tables

The result, REVIEW_TOPICS, kept up to date

Dynamic tables

A target lag sets how fresh the result must be

Dynamic tables

177 labeled reviews are left as they are, and AI_CLASSIFY runs once for the new review

Ask for the outcome

The request for a review topics table, in a new schema called PIPELINE

Ask for the outcome

The request, with one topic for each review highlighted

Ask for the outcome

The request, with staying up to date within two minutes highlighted

Review before approving

The Execute SQL box, with DYNAMIC TABLE highlighted

Review before approving

The Execute SQL box, with the two-minute target lag highlighted

Review before approving

The Execute SQL box, with AI_CLASSIFY and the five topics highlighted

Review before approving

The Execute SQL box, with the reviews table highlighted

A new review arrives

A new two-star review of the USB-C Hub lands in SHOP.REVIEWS

A new review arrives

We ask CoCo whether REVIEW_TOPICS has labeled the new review

A new review arrives

CoCo shows review 178 in REVIEW_TOPICS with the topic delivery

Check the refresh history

We ask CoCo for the refresh history of REVIEW_TOPICS

Check the refresh history

The first refresh, at creation, inserted 177 rows

Check the refresh history

A scheduled refresh found no data and inserted nothing

Check the refresh history

A scheduled refresh inserted one row, the new review

A scheduled task

The request for a weekly summary of reviews

A scheduled task

The request, with every Monday at 9am New York time highlighted

A scheduled task

The request, with a two-sentence summary of the past week highlighted

Inside the task

The Execute SQL box for the task

Inside the task

The Execute SQL box, with the cron schedule highlighted and its fields explained

Inside the task

The Execute SQL box, with AI_AGG highlighted

Turn the task on

The Execute SQL box to resume the task, with RESUME highlighted

Test it once

We ask CoCo to run the task once and show the run history and the saved row

Test it once

The run history, with the successful run highlighted

Test it once

The saved summary, highlighted

Pause when needed

We ask CoCo to pause REVIEW_TOPICS and the weekly task

Pause when needed

Two Execute SQL boxes, with SUSPEND highlighted in each

CoCo inside the pipeline

CoCo builds the pipeline, then Snowflake runs it

CoCo inside the pipeline

CoCo itself runs inside the pipeline

Automations

The cortex automation create command

Automations

The command, with the name highlighted

Automations

The command, with the prompt highlighted

Automations

The command, with the weekday schedule highlighted

A report in your inbox

The automation report email

A report in your inbox

The email subject, highlighted

A report in your inbox

The two flagged products in the email body, highlighted

CoCo in GitHub Actions

GitHub Actions runs jobs on your repository automatically, for example each time the code changes

Snowflake example workflow

Snowflake example GitHub Actions workflow, a YAML file

¹ https://github.com/snowflakedb/snowflake-actions

Snowflake example workflow

The workflow, with the cortex exec step highlighted

¹ https://github.com/snowflakedb/snowflake-actions

The full picture

Dynamic tables keep a table current as data arrives

The full picture

Tasks run SQL at a fixed time

The full picture

Automations and GitHub Actions run CoCo itself

Let's practice!

Introduction to Snowflake CoCo CLI

Preparing Video For Download...