AI agents
AI agents that do the work inside your systems
An agent is not a chat window. It reads the enquiry, checks your data and creates the CRM record, the supplier order or the reply to the client.
When you need one
- Enquiries arrive at night and at weekends, and the manager replies in the morning.
- Clients write in different languages.
- Every morning someone counts stock and builds supplier orders.
How it works: a sales example
- An enquiry arrives in WhatsApp.
- The agent replies in the client's language.
- It asks about goal, budget and timing.
- It creates the CRM card.
- The manager gets an alert and a ready card.
What the agent does and does not do
| Does on its own | Only after a person says "ok" |
|---|---|
| Answers routine questions | Sends the order to a supplier |
| Qualifies the lead | Discounts and exceptions |
| Fills in the CRM | Anything involving money |
| Forecasts stock | Unusual cases: hands over to the manager |
Systems
WhatsApp Business API, Telegram, Instagram, HubSpot, Pipedrive, Kommo, POS systems, Google Workspace. Models: Claude and Gemini, with RAG on your data.
Timeline
One process usually takes 2–4 weeks. The agent first runs in shadow mode next to your team, then goes live.
Data
Everything runs in your accounts, under GDPR and a DPA, with every step logged.
Price
Audit from €500, credited towards implementation · implementation from €1,500 · support from €150 a month.
Case studies
Questions
Can the agent make mistakes?
Yes. That is why money and exceptions go through a person, and every step is logged.
How is this different from a $15 chatbot?
A bot follows a script. An agent works with your data and writes into your systems. If a bot is enough, the audit will say so.
Who owns the accounts and data?
You do. Access and documentation stay with you.