Smart Import AI

Features

Smart Import is the fastest way to get an org chart out of whatever data you already have. Drop in a messy spreadsheet, a Notion export, a Slack message describing your team, or just paste an email — the AI reads it, extracts the people and reporting structure, and builds the chart for you.

You don't have to clean up your data first. You don't have to rename columns. You don't even have to give it a CSV.

What it accepts

• CSV files — from Excel, Google Sheets, BambooHR, Gusto, Workday, Rippling, or any HR system. Column names can be anything (Name/FullName/Employee, Title/Role/Position, Manager/ReportsTo/Boss).
• TSV / tab-separated — same flexibility, different delimiter.
• Pasted text — copy-paste a list, an email, a Slack message, or even a few sentences describing your team. The AI figures out the structure.
• Plain JSON — for technical users, structured exports work too.

How to use it

1. Open OrgCanvas and click the File menu in the toolbar
2. Click Import a File AI (the option with the sparkle icon)
3. Drop your file or paste text into the box
4. Click Extract people
5. Wait 3–8 seconds while the AI reads your data (longer for free-form prose)
6. Review the preview — counts of people, departments, and any items the AI flagged
7. Click Add to chart (or Replace existing if your chart already has people)

What it figures out automatically

Smart Import doesn't just match columns — it actively resolves common data problems for you:

• Duplicates — Bob O'Neill and bob oneill get merged into one person.
• Multi-manager rows — if a row says "Alice / Bob" or "Alice, Bob" in the manager field, the first becomes the primary manager and the second becomes a secondary (dotted-line) reporting line.
• Department normalization — Eng, Engineering, and engineering collapse to a single canonical department.
• Co-leadership — if the data shows two co-CEOs or co-founders, the AI flags them and asks if you want them rendered as peers (instead of one above the other). You can opt in or out per pair.
• Title-as-top-level — if someone's manager is listed as a title like “CEO” with no actual person matching, that person becomes a top-level root instead of orphaning them.
• Encoding glitches — stray Unicode replacement characters (the � you sometimes see in copy-pasted text) get stripped before matching.

The review panel

Before anything is added to your chart, Smart Import shows you a preview with a few things to confirm:

• Summary — number of people, departments, and reporting relationships extracted
• Co-leadership detected — opt-out checkboxes for any co-leader pairs the AI inferred (uncheck to render them in a normal hierarchy instead)
• Things to review — ambiguous rows the AI couldn't fully resolve. Each one shows what the AI guessed and what the alternative would be.
• Auto-resolutions — a transparent list of dedupes, normalizations, and other changes the AI applied so you know exactly what happened to your data.

If your chart already has people

When you import into a chart that's not empty, you'll get a 3-way prompt:

• Replace — wipe the existing chart and use the imported people instead. Use this when you're starting over or replacing a stale snapshot.
• Add to chart — merge the imported people into what you already have. Useful when you're adding a new team or division.
• Cancel — back out without changing anything.

What happens to people without a manager

If the AI can't figure out who someone reports to (e.g., the manager field is blank or names someone not in the import), that person is added to the chart but doesn't appear on the visual hierarchy. Instead they show up in the Unassigned section of Table view, and a small chip on the canvas tells you how many people need attention.

This is intentional — floating cards make a chart look broken. Park them in the Roster, then assign them when you're ready.

How it compares to classic CSV import

The legacy Import CSV (classic) option is still available in the File menu. It's a strict column-mapped import: your CSV needs specific column names (name, title, manager, etc.), and it doesn't dedupe, normalize, or handle prose. Use it if you want exact, predictable behavior with a clean CSV that already matches the expected format.

For everything else — messy data, weird column names, free-form text, or anything that's not a perfectly-shaped CSV — use Smart Import.

Troubleshooting

• “We couldn't find any people in that file.” — The file probably doesn't contain recognizable name/title columns or descriptive text. Try pasting a simpler list with one person per line.
• Import is taking longer than 8 seconds. — Long prose or very large files take longer. The system automatically falls back to a faster model if the primary one times out.
• The AI got something wrong. — Use the review panel to opt out of co-leadership detection or fix individual rows after import via Table view. You can also undo the entire import with Cmd+Z.