The problem with generic chatbots
A generic chatbot is a bit like a salesperson who doesn't know your industry. They respond, sure, but beside the point. They use vague terms, offer unsuitable solutions, and end up frustrating the customer rather than helping.
According to Gartner, 64% of customers prefer not to interact with a chatbot at all rather than use one that doesn't understand their problem. Yet companies continue to deploy copy-paste solutions, hoping that volume will make up for lack of relevance.
What personalisation really means
Personalisation isn't just about displaying the customer's first name or changing the button colour. It's about deep domain knowledge built directly into the AI model.
An H'appi chatbot for a logistics carrier knows:
- Industry terminology (BL, CMR, ETD, ETA, INCOTERMS)
- Typical delivery and claims processes
- Common TMS and WMS integrations
- Appropriate legal and contractual responses for each type of dispute
Our personalisation process
Every H'appi project starts with a scoping workshop lasting one to two weeks. We document:
- Your customers' 20 to 30 most frequent requests
- Vocabulary specific to your business
- Edge cases and mandatory escalations to a human
- The desired tone of voice (formal, casual, technical)
What it changes in practice
The difference between a generic chatbot and a domain-specific one is measurable:
- First contact resolution rate: a generic bot sits around 30-40%. A well-trained domain bot can exceed 70%.
- Escalation rate: fewer transfers to human agents = lower costs, less waiting.
- Customer satisfaction: users immediately sense they're talking to something that understands them.
Our honest position
H'appi is a young startup. We don't claim to have hundreds of client cases. What we have is a rigorous method, founders who have worked on AI projects in enterprise settings, and the conviction that personalisation is non-negotiable.