Data Mapping and Transformation Without Code

If you need to take raw, messy files and turn them into clean, standardized data — without writing code — Obvious handles the full workflow. You can import files, map columns, apply transformations, flag bad data, and save the entire process as a reusable template so it repeats reliably with new files.

Start with your goal

Before importing anything, start a new project and tell Obvious what you're trying to accomplish. Something like:

I need to clean up a messy customer export and standardize it for our CRM.

This context helps the agent stay oriented as you work through each step — it knows what the final result should look like, not just what the next action is.

Import your data

Obvious accepts CSV, Excel (.xlsx, .xls), and JSON files. To bring data in, drag a file directly into the chat, use the + button at the bottom-left of the chat input and select Upload File, or just describe what you want and let the agent handle it.

When you upload a spreadsheet, Obvious reads your column headers, detects field types (text, numbers, dates, checkboxes), and creates a workbook with your data ready to go.

For more detail, see Importing Data.

Map columns to your standard

If your incoming files use different column names than your target format — for example, a file that says "First Name" where you need "given_name" — you can ask the agent to remap them.

Import this file and map the columns to match my existing sheet.

The agent handles inexact matches intelligently, even when names differ in capitalization, spacing, or phrasing. You can describe the mapping you want in plain language, and the agent applies it.

If you're importing into a sheet that already has columns set up, Obvious matches file headers to existing field names automatically. Exact matches map on their own; anything that doesn't match gets added as a new column or flagged for your review.

Transform your data

Once your data is in a sheet, you can ask the agent to transform it in plain language. Common examples:

• Normalize values: "Standardize the status column so all values are lowercase."
• Split or combine fields: "Split the full name column into first name and last name."
• Reformat dates: "Convert all dates in the order date column to YYYY-MM-DD format."
• Deduplicate: "Remove any rows with duplicate email addresses."
• Reclassify: "Set the category to 'Unknown' for any row where it's blank."

The agent applies the transformation directly to your sheet. Take your time getting the data exactly how you want it — you can review the results and ask for adjustments in the same conversation.

For AI-powered column generation — where Obvious fills a new column based on patterns in your existing data — see Enrichments.

Validate your data

Validation rules check every row against conditions you define and flag anything that doesn't pass. You can add them by asking the agent:

Flag rows with missing email addresses. Flag any phone number that doesn't look valid. Flag order amounts below 0 or above 10,000.

Cells that fail show color-coded indicators — red for errors, yellow for warnings, blue for informational flags. Hover over any flagged cell to see exactly what's wrong.

Validations run automatically on existing data and catch issues as new data arrives. For the full list of validation types, see Validations.

Save the workflow as a reusable template

Once you've worked through the import, mapping, transformation, and validation steps and the data looks right, you can save the whole process so it's ready for next time.

First, ask the agent to document what you did:

Write a step-by-step guide of everything we did in this project so I can repeat it.

The agent creates a document in your project capturing each step — the file you imported, the mappings you applied, the transformations you made, and the validations you set up.

Then save the project as a template:

Save this project as a workspace template called "Data Cleanup Workflow."

The template packages your project's artifacts, files, and structure into a reusable starting point. Next time you have a similar file, create a new project from that template and the agent will have the full workflow context ready to go. For step-by-step instructions, see Create a Template in the UI.

If you'd rather fully automate the process — so it runs on a schedule or triggers automatically when a new file arrives — you can save it as a task instead. Ask the agent to create a task from the steps you've built. Tasks can include approval steps if you want a human to review the output before it moves forward.