Set up your business with AI

pro

Use the conversational builder (or a spreadsheet) to create real sites, staff, services and customers.

The Set up my real business with AI tool turns a short conversation — plus any spreadsheets you already keep — into real, live records in your organization: locations, your team, your service catalogue, inventory, and customers. Unlike Load example data, which loads a fictional demo template you can explore and discard, this builder writes the actual data you'll run your business on.

You'll find it on the Configuration page as the Set up my real business with AI card, which opens the builder at /configuration/business-setup/.

Everything you confirm here is written to your live organization — not a sandbox. Created services and inventory immediately feed bookings, sales, and analytics; customers become real CRM contacts. Always check the review screen before confirming. There is no one-click undo for a manager.

Who can do this

The builder bulk-creates real data, so it's gated to managers and the platform operator.

Action	Who can do it
Open the builder, chat, upload files, review and build	Domain manager (`is_manager`)
Replace existing data first (the optional reset before building)	Apolaris (the platform operator) only — never a manager

A domain manager always passes the gate. A regular employee or a privilege admin (can_manage_privileges without is_manager) cannot open the flow. The replace existing data first option is reserved for the Apolaris AI platform operator; managers never see it.

How it works

The builder collects a setup — a structured plan of what to create — across the conversation. Nothing touches the database until you reach the review step and confirm.

Open Configuration → Set up my real business with AI and start the builder.
Chat with the AI onboarding consultant. It asks focused questions about your locations, team and roles, service/product catalogue with prices, inventory, and customers, and keeps a running side panel of what's been captured.
Upload spreadsheets (CSV or Excel) of customers, products, services, or staff. The importer reads each file, classifies it from its column headers, and folds the rows into the setup. The side panel updates with a summary of what was imported.
Review a read-only preview of exactly what will be created.
Build — after you tick the confirmation box, the real records are written to your organization.

You can start over at any time before building (this clears the captured setup but writes nothing), or cancel to leave the flow.

The builder has no external integrations — everything is captured directly in-app from your chat answers and uploaded files. It also degrades gracefully: if AI is turned off, the conversation falls back to simple acknowledgements, and file import (the bulk of the value) still works, so you can drive the whole build by uploading your data.

Uploading spreadsheets

This is the fastest path if you already keep lists of your customers, products, services, or staff. Attach one or more files to a message and the importer does the rest.

Supported formats: CSV and Excel (.xlsx).
The system guesses the file type from its headers. Columns are matched by name in English, Spanish, and French — for example, a sheet with email and phone (or correo and telefono) plus a name column is read as customers; a sheet with salary/position and a name is read as staff; service + price + duration is services; sku/product/stock/cost is inventory (products).
Every sheet of a workbook is read. Each tab is classified on its own, so a file with a services tab, an extras tab and a products tab imports them all in one go. (Very large files stream in the background and read the first sheet only — split tabs into separate files there.)
Title rows are fine. A title or blank rows above the real header row (a very common shape for exported sheets) are skipped automatically — the importer finds the actual column headers on its own.
If a sheet still can't be classified, the assistant asks what it holds. Just answer — "they're services", "son clientes" — and the file you already uploaded is imported right away as that type; no re-upload needed.

A clean header row is what makes import accurate. Common column names work out of the box: First Name/Last Name or a single Name, Email, Phone, Position, Salary, Service/Product, Price, Cost, Duration (minutes), Category, SKU, and Stock.

Limits for in-chat uploads

Small files are read inline, right inside the chat — fast and conversational. These ceilings keep that in-chat path responsive. A bigger file isn't rejected: it takes the background import path instead (next section), so nothing is skipped.

Limit	Value
Read inline in chat	Files up to about 1.5 MB and 3,000 rows
Files per message	6
Sites (locations)	20
Departments	40
Roles	40
Total team headcount (named staff + role counts)	300
Services	400
Inventory items	600
Customers	3,000

A file larger than about 1.5 MB, or with more than 3,000 rows, is streamed in the background instead of read inline (see the next section). If a list exceeds its cap, the extra rows are dropped. Duplicate rows — by email or name+phone for people, by category+name for products, by name for services and sites — are merged automatically.

Large files import in the background

A big customer or product list no longer has to be trimmed to fit the chat. When a file is larger than about 1.5 MB or has more than 3,000 rows, Apolaris hands it to a background import instead of reading it inline — so the page never freezes and even very large lists load.

Here's what changes for big files:

It streams. The file is read in batches and written into your organization a chunk at a time, so memory stays bounded no matter how large the file is.
You see live progress. A Creation Tasks panel shows each import's status — rows processed, records created and skipped, a percentage, and an estimated time remaining — and it keeps updating even if you reload the page or leave and come back.
You can cancel. Stop an import mid-run; everything already imported is kept (the task finishes as cancelled, not rolled back).
You're notified when it's done. You get an in-app notification (and a browser push, if you've allowed it) when an import finishes, deep-linked back to the builder.

Finish states you may see: done (everything imported cleanly), partial (some rows were duplicates or skipped — the file is kept so you can re-run it), cancelled (you stopped it; imported rows are kept), or failed (an error stopped it).

The background path accepts files up to 100 MB each. Split anything larger across several files — each becomes its own import task, and the system runs a couple per organization at a time and queues the rest. These streamed imports go straight into your data and aren't bound by the in-chat session caps above (those exist only to keep the conversational build snappy).

A background import is additive and starts as soon as you upload the file — unlike the conversational build, it does not wait for the review-and-confirm step, and there is no one-click undo. Duplicate rows are merged automatically, but new records go straight into your live organization. Apolaris works out what a large file contains from its header row, so a clean, clearly-labelled header keeps every row routed to the right place — and if it can't tell what a big file is, it asks rather than guessing.

The review step

Before anything is written, the review screen (/configuration/business-setup/review/) shows a display-ready preview: your business profile, every site, department, role, named staff member, service and inventory item, plus a sample of customers and the total headcount a build would create. It also flags whether the setup is "ready to build."

A setup is considered ready once it has at least one site and at least one service or inventory item. The screen still lists anything recommended but missing — your business name, a team (staff or roles), or a customer list — so you can go back and add it before confirming.

Read the headcount and counts carefully. Roles expand into individual employees — a role with a count of 50 creates 50 staff records. Numbers, prices, and durations are taken from your files and answers; fix any wrong values in your source data and re-upload rather than building twice.

What a build creates

When you confirm, the builder writes records onto your organization in a single batch:

Sites (locations) — created if a site of that name doesn't already exist.
Departments — for organizing staff.
Team — your named staff first, then one employee per role count. Each employee is linked to your sites; a recognized position is matched to a department.
Services — added to your catalogue (existing names by the same name are left untouched). A service with no duration defaults to 60 minutes.
Inventory — products with their categories, plus per-site stock (opening stock lands at your first site).
Customers — with a required, unique email per organization. If a customer's email is blank or collides with another, the builder synthesizes a placeholder address so the import never fails.

The build is additive and conservative. It uses get-or-create on sites, departments, and services, so re-running it won't duplicate things that already match by name. It will not rename your organization or change a currency that's already in use, and it only stamps an industry if it matches a recognized preset.

Cross-data impact: the records you create are immediately live. New services and products can be sold and booked right away, sales and stock movements against them flow into your reports, and customers appear in search, the calendar, and analytics. Plan to build once with reviewed data rather than iterating against your live org.

Replace existing data first (platform operator only)

When the Apolaris AI platform operator runs the builder, the review screen offers a replace existing data first option. If selected, it wipes the organization's business data — sales, bookings, expenses (and budgets/recurring expenses), inventory, customers, services, employees, and sites — before writing the new setup. User logins and account settings are kept.

This reset is reserved for Apolaris by design and a manager will never see or be able to run it.

The reset is destructive and not reversible. Because it deletes transactional records (sales, bookings, expenses), it desyncs historical analytics and reports for the period it removes. It exists for clean re-loads, not routine cleanup — if you think you need it, contact support rather than trying to reset live data yourself.

Set up with AI vs. Load example data

Both tools live on the Configuration page and are gated to managers, but they do very different things.

	Set up my real business with AI	Load example data
Creates	Your real records	Demo / sample data
Input	Chat answers + CSV/XLSX uploads	Pick an industry vertical + AI suggestions
Best for	Going live with your own data	Exploring, training, demos
URL	`/configuration/business-setup/`	`/configuration/starter-data/`

If you want to see a fully populated CRM before committing your own data, start with Load example data. When you're ready to run your actual business, use this builder with your own spreadsheets. See Onboarding & industry presets for the example-data wizard.