Oleon

AI agent for WhatsApp

An agent that answers, and knows when not to.

The problem with a bot is rarely that it cannot talk. It is that it will not stop talking when it does not know. An agent that invents a delivery date costs more than the salary it saved.

Grounded, not improvised.

Answers come from a knowledge base you fill in. Asked something outside it, the agent says so and hands over rather than guessing.

It hands over to a person.

The handover is a first-class thing, not a fallback. The conversation moves into the shared inbox with its history intact, so whoever picks it up is not asking the customer to start again.

It looks things up mid-conversation.

An order status or a booking is fetched while the customer waits, rather than answered from a script written last quarter.

Three languages, properly.

Sinhala, Tamil and English, which for most Sri Lankan businesses is not a nice-to-have but the actual shape of the inbox.


Tech specs

Every one of these is a screen in the product.

  • Grounded answers
  • A knowledge base
  • Greetings
  • Auto replies
  • Follow-up messages
  • Looks things up mid-chat
  • Hands over to a person
  • Sinhala, Tamil and English

Questions

Before you ask.

The ones we hear most about AI agent for WhatsApp. If yours is not here, the answer is a message away.

What happens when the agent does not know the answer?

It says so and hands the conversation to a person, carrying the history with it. It is built not to fill a gap with an invented answer.

Do we have to train it?

You fill in a knowledge base. There is no model training step and no dataset to prepare.

Can it answer in Sinhala and Tamil?

Yes, in Sinhala, Tamil and English.


It starts with one conversation.

Bring the channels you already answer on into one place, and see what the rest of the week looks like.

Not ready to sign up? Join the waitlist and we will come to you.

AI agent for WhatsApp · Oleon Workspace