Voice to Text for Data Engineers
Your brain holds the context of complex pipelines and transformations. Writing it down shouldn't slow you down. Blurt lets you speak your documentation, SQL comments, and Slack updates while your hands stay on the keyboard. Hold a button, explain your data flow, release. Text appears wherever your cursor is — in Jupyter, VS Code, Snowflake, Airflow, anywhere. No more choosing between building pipelines and documenting them. Just talk and type.
The Typing Problem
Documenting pipeline logic while it's still fresh
You just spent two hours building a complex DAG with tricky edge cases and conditional branching. You know exactly why each step exists right now. But by tomorrow, you'll have moved on to three other tasks. The documentation never gets written. Six months later, you're staring at your own code wondering what past-you was thinking.
SQL comments that explain the why, not just the what
Your query joins five tables with a filter that looks arbitrary but actually handles a gnarly data quality issue. You should document why. But after writing 200 lines of SQL, typing out the explanation feels exhausting. So you add a TODO comment and move on. That TODO has been there for eight months now.
Data dictionary updates that never happen
The new column needs documentation. The stakeholders need to understand what 'is_active_v2' means. You know exactly what it means — you could explain it in 15 seconds out loud. But navigating to the wiki, finding the right page, and typing it out takes 10 minutes you don't have before the next meeting. The dictionary stays incomplete.
Slack questions during deep pipeline debugging
You're tracing a data quality issue through four transformation stages when an analyst pings you about a dashboard discrepancy. You could answer quickly — it's the same root cause — but typing the explanation means losing your mental map of the pipeline. By the time you've typed your reply, you've forgotten which stage you were investigating.
Your hands ache from typing SQL all day
Hundreds of lines of SQL. Pipeline configurations. Slack messages. Documentation. Jira tickets. By Wednesday afternoon, your wrists are protesting. You've tried the ergonomic keyboard, the standing desk, the wrist rests. But you're still typing thousands of words every day. Your hands weren't designed for this volume.
How It Works
Blurt works in every tool data engineers use — Jupyter notebooks, VS Code, Snowflake console, Airflow UI, Slack, Confluence. Anywhere you can type.
Hold your hotkey
Press your chosen shortcut. A small indicator shows Blurt is listening.
Explain naturally
Say your pipeline description, SQL comment, or data definition. Blurt adds punctuation.
Release and done
Text appears at your cursor. No copying, no pasting, no extra clicks.
Real Scenarios
Documenting ETL jobs before you forget the edge cases
You just finished configuring an Airflow DAG with retry logic for a flaky API. The edge cases are fresh in your mind. Hold your hotkey and explain: 'This DAG handles the nightly customer sync. It retries three times on API timeout because the vendor rate limits after 10 PM. The backfill task runs separately to avoid hitting quotas.' Documentation done in 15 seconds while the context is still in your head.
Adding SQL comments that actually help
Your CTE has a filter that looks weird but prevents a subtle data duplication issue. Cursor above the line, hold button, say 'This filter excludes records where the source system sent duplicate events during the March migration. See incident report JIRA-4521 for details.' Future maintainers will actually understand the code. 10 seconds, not 2 minutes of typing.
Explaining schema changes to stakeholders
The analytics team is confused about the new column you added. Instead of typing a long Slack message, hold and talk: 'The customer_lifetime_value column calculates total revenue from order_items, excluding refunds and test transactions. It updates daily at 3 AM UTC. Values are in cents to avoid floating point issues.' Reply sent. No typing required. The analyst has what they need.
Writing data dictionary entries on the fly
You're adding a new table and the data catalog needs updating. Hold button, describe the table: 'The user_sessions_daily table aggregates raw clickstream events into daily session summaries. One row per user per day. Partitioned by event_date for query performance. Refreshed nightly by the sessions_aggregation DAG.' Dictionary entry complete without context switching to a wiki editor.
Responding to pipeline alert investigations
The on-call channel is pinging you about a failed job. You've already diagnosed the issue. Hold and explain: 'The warehouse_sync job failed because Snowflake maintenance ran long last night. The data landed 2 hours late but is complete. I've already triggered a backfill for the downstream models.' Team is informed. You're back to fixing the actual problem.
Updating Jira tickets with technical context
Your ticket needs implementation details before sprint review. You know exactly what you did but typing it out takes forever. Hold button: 'Implemented incremental load pattern using last_modified timestamps. Added monitoring for row count drift. Performance improved from 45 minutes to 8 minutes due to partition pruning.' Ticket updated in seconds. The scrum master actually knows what you built.
Jupyter notebook markdown cells
Your analysis notebook needs context for non-technical reviewers. Instead of typing markdown, hold and explain: 'This section calculates customer churn rates by cohort. The sudden spike in cohort 2023-Q2 corresponds to the pricing change that month. See the retention analysis notebook for deeper investigation.' Your notebook tells a story. Stakeholders can follow along.
Why data engineers choose Blurt over built-in dictation
| Blurt | macOS Dictation | |
|---|---|---|
| Activation | Single hotkey, instant start | Click microphone icon or fumble with settings |
| Speed | Text appears in under 500ms | 2-3 second delay before transcription begins |
| Technical terms | Handles SQL keywords and data terms well | Struggles with words like 'nullable', 'partitioned', 'backfill' |
| Reliability | Consistent accuracy session after session | Often fails silently or produces garbled output |
Frequently Asked Questions
Start Typing Faster Today
Free to try — no credit card required
Download Blurt