Skip to content

Search documentation

Find a guide in the Raileon documentation

How your agents learn

Last updated

Your agents are not static. They get better over time through three layers of learning.

1. Onboarding data

Everything you share with us on your audit call and during setup goes into your agents' private configuration. This is your agents' starting point: your industry, your workflows, your tools, and your goals.

2. Documents you upload

Brand guidelines, email templates, standard procedures, customer lists, and product details all get added to the knowledge base. Your agents reference these before they act, so their output matches how your business actually works.

3. Feedback from your decisions

Every time you approve or decline a suggested action in your Approvals drawer, that decision is signal. When the same kind of correction shows up repeatedly — declining the same tone, pushing back on the same kind of claim — your agent turns the pattern into a standing preference, so it stops repeating it without you having to say so again. This review runs automatically, about once a day.

It compounds

The fastest way to improve your agents is to upload good examples and to give quick feedback in the first few weeks. That early input pays off for months.

Where to go next