The AI agent is the core of Billabex. Acting as your virtual teammate, it manages invoice follow-ups and collections with one goal: maximize recovery while preserving your customer relationships.
What Exactly Does the AI Agent Do?
As soon as an invoice is imported or synced, the AI agent takes over. It:
1. Analyzes each invoice and credit note — amount, due date, payment status, customer history
2. Builds a smart follow-up plan — tailored to the client’s profile, tone of voice, and situation (age of debt, disputes, promises, etc.)
3. Communicates automatically — via email or letter depending on available data (soon: SMS and voice messages)
4. Tracks replies — and adjusts behavior in case of payment promises, disputes, or custom requests
5. Collaborates with your team — by requesting input only when human intervention is necessary (via tasks)
Invoice Lifecycle
The follow-up plan is dynamic. At each stage, the AI agent:
- Checks the latest payment status
- Reprograms follow-ups based on aging
- Adjusts the tone (courteous → professional → firm)
- Stops actions upon closure or manual override

All communications are scheduled on business days, adjusted based on the client’s country.
Email Deliverability Tracking
The AI agent includes automatic deliverability tracking to ensure messages reach their recipients.
If an email fails (e.g., invalid address, hard bounce), the agent:
- Marks the contact as invalid on their account
- Creates a task asking the team to provide a valid contact
- Pauses follow-ups until the contact is updated
- Tracks the issue via the Tasks module

This protects the sender reputation and ensures future follow-ups remain effective.
Communication Language
The AI agent automatically adapts the language used in all messages — whether addressing client contacts or your team members.
Language is determined by:
1. The language defined at the contact level
2. If not defined, it is inferred from the contact’s country

No configuration needed — clients receive messages in their own language.
For your team
When communicating with internal users (e.g., task notifications), the agent applies the following logic:
1. If a user has a preferred language set, it is used
2. If not, the default language is English
When several users are recipients:
- If they all share the same preferred language → that language is used
- If preferences vary → the agent uses the majority language, or defaults to English if unclear
When the AI Agent Reaches Out to You
Although autonomous, the AI agent requests your input in cases like:
- Ambiguous customer replies or disputes
- Missing or inconsistent data (e.g., no contact available)
- Detected issues (e.g., explicit refusal, invalid email, callback request, no reply after 4 messages)
- Requests from your internal team
In these cases, a task is automatically created to ensure follow-up.

You’ll also be notified of good news — such as a payment promise or when proof of payment is submitted.
Does the AI Agent Learn?
The AI agent is not self-learning but is powered by a contextual intelligence engine. It:
- Adapts follow-ups based on past interactions
- Avoids re-contacting clients already in conversation
- Respects your organizational and language preferences
- Applies specific rules like email validity or due dates
- Considers instructions from contact and internal notes
Regular updates ensure ongoing improvement without affecting stability.