Hallucination
An AI model producing a confident but unsupported answer. In customer messaging it looks like an invented policy, price or delivery promise.
One term from the WhatsApp Business Platform glossary. Last updated 2026-07-25.
Language models generate plausible text, and plausible is not the same as correct. Asked about a refund window it has no information on, an ungrounded model will often produce a reasonable-sounding number rather than admitting uncertainty.
Grounding through retrieval reduces this substantially but does not eliminate it, which is why scenario evaluation before launch and explicit handoff rules in production both remain necessary. A commitment made by an agent is still a commitment your business made.
Related terms
- Retrieval-augmented generation (RAG)
- A technique where an AI agent retrieves passages from your own knowledge base before answering, so replies are grounded in your documented policies and product facts rather than invented.
- Knowledge base
- The collection of business documents an AI agent retrieves from — policies, product information, FAQs — so its answers reflect your business rather than general internet knowledge.
- Human handoff
- The transfer of a conversation from an AI agent to a person, triggered by topic, repeated failure to resolve, negative sentiment or an explicit customer request.
See Hallucination in a real workspace
BotPulsar runs the official WhatsApp Business Platform with unlimited team members on every plan, starting at ₹300/month.