Watbot: the platform for bots and AI agents my partner and I have been building since 2018
Watbot is the platform businesses use to build bots and AI agents without a developer. It has been running since 2018 and more than 70,000 bots have been built on it. It is a project my partner and I run together, so I know it from the inside and build client projects on it myself.
An agent with your knowledge base
Your services, terms and answers to common questions go into a knowledge base. The agent answers from it, and hands anything that is not in it to a person along with the conversation.
The channels your customers already use
Telegram, WhatsApp and the chat on your site. One agent and one knowledge base across all of them, rather than answering in each separately.
Built for you, end to end
Conversation design, knowledge base, integrations, launch and support. You describe the task, I build it, and you see the agent working before it goes live.
The account stays yours
It is opened in your name, and the knowledge base and conversations stay with you. After launch you can run the agent yourself or leave the upkeep with me.
My part in it
Watbot is a project my partner and I run together, and I am its co-founder. The platform has been running since 2018, and more than 70,000 bots have been built on it in that time.
For you that means something practical. When I build an agent on Watbot I am not learning someone else’s tool from its documentation: I know what the platform does, what it does not, and what can be changed in it to fit a job. If a task runs into the platform itself, that is a short conversation for me.
What Watbot is
A builder where a business puts together a bot or an AI agent for talking to customers, without hiring a developer. The agent draws its answers from a knowledge base you fill with your own material, and works in the channels your customers already write to.
The difference between a “bot” and an “AI agent” here is a practical one. A bot follows buttons and a script written in advance. An agent understands a question put in ordinary words and answers from the knowledge base. Both can be built on the same platform, and in practice the thing that works is usually a mix: a fixed script where a step is compulsory, a free answer where people ask in their own words.
When it fits
When you have a stream of similar questions and the answers already exist somewhere in your documents, price lists and past replies. The knowledge base then comes together quickly, and the agent takes the repetitive part off a person.
When every question is different and needs a decision rather than a fact, an agent gives you less. It is more honest in that case to use it only for intake: find out what is needed and hand it to a person with a ready summary.
What happens next
Tell me which questions you get most often, and where. I will say which of them an agent closes and which it does not, and propose a shape for it. There is finished work on the platform in the case about the knowledge base plugin, and the platform itself is at watbot.org.