Teach the AI your language. Not the other way around.
Training means helping the AI understand what your data means. You name the rules. It learns.
Before / after
Without your labels, answers sound generic. After a short training pass, the same question returns numbers and terms from your columns — because the model knows what they mean.
Before
"Revenue looks positive based on typical patterns."
After training
"Q3 net revenue was $1.2M per your P&L tab — up 8% vs Q2."
Three starters
Business finance
P&L analysis, AR aging, revenue trends.
- What's driving margin this quarter?
- Which invoices are past due?
- Compare revenue to last year
Health & wellness
HRV, sleep scores, activity patterns.
- How did sleep affect HRV?
- Weekly recovery trend
- Compare this month to last
Investment portfolio
Holdings, returns, risk exposure.
- Largest position by weight?
- Realized vs unrealized
- Sector concentration
Model builder
- 1
Choose a template or start blank
- 2
Connect your uploaded data
- 3
Label your key columns
- 4
Ask test questions and refine
- 5
Save and name your model
FAQ
How long does training take?
Most first models finish in minutes. Larger files add time — we show progress the whole way.
Do I need to retrain when my data changes?
Upload a new version and refresh — the model can incrementally learn what changed.
Can I have multiple models?
Yes — one per use case, dataset, or team. Plans scale with how many you need.
What if my data has errors?
We flag anomalies during upload. You fix rows or exclude columns before training.
Can I share a model with my team?
Business plans include seats so colleagues use the same trained model securely.
What does "labeling" mean?
Telling the AI which columns are dates, money, categories, or IDs — so answers match how you think.