Build conversational AI for applications.

What is Artificial Intelligence as a Service?

A RESTful API for integrating artificially intelligent chatbots into applications.


Control devices via applications using natural language commands, e.g.,
"Turn off the lights!"

Customer Care

Reduce support queries, monitor forums, and automate tasks using a chatbot that knows your company's FAQ.

Scripting Engine

Create interactive content for advertising or entertainment applications like brand ambassadors or talking characters.

Messaging Applications

Add a bot to your Slack channel, or integrate with services like Twitter, Hubot, Facebook, Skype, Twilio, and more.

Virtual Assistance

Build your own Siri, or something more useful like a hands-free assistant for driving, cooking, or exercising.

Domain Experts

Craft knowledge bases for specific verticals like healthcare, financial services, or education.


Select the Developer Plan to start your free trial


$ USD 9.00 /month
  • Open Source Libraries


$ USD 99.00 /month
  • Open Source Libraries
  • Email Support


$ USD 999.00 /month
  • Open Source Libraries
  • + Phone Support


  • Unlimited Interactions
  • Bot Network Design
  • Premium Libraries
  • + Engineering Support
  • Database Integration
  • Service Level Agreement
  • More Custom Features

Call 415.343.5894 or email to ask about Enterprise-level features and tools.

Getting Started

Sign up and install the command line interface to start playing with the API.


The CLI requires node.js. If installed, download the CLI using npm.

$ npm install -g pb-cli


Store your API credentials in a file called chatbot.json. The init command helps create the file by prompting you for the required information.

$ mkdir alice
$ cd alice
$ pb init
$ pb create


The current offering supports Artificial Intelligence Markup Language (AIML), the open standard for writing chatbots. AIML is an extensible XML dialect with powerful dialog-scripting and natural language processing capabilities.

Using your favorite text editor or the Playground IDE, create a new AIML file.

<?xml version="1.0" encoding="utf-8" ?>
<aiml version="2.0">

<template>Hello world!</template>



Upload files to the cloud, and compile to update and check for any syntax errors.

$ pb upload example.aiml
$ pb compile


Test out your bot by talking to it.

$ pb talk Hi

Powered by Pandorabots

I’m using Pandorabots to power the software brain in Elias, a voice-controlled hardware robot made with MyRobotLab & InMoov. Using AIML features like the <learn> and <oob> tags, I can speak to and control Elias through a series of commands and can even teach him new behaviors. AIML also allowed us to reduce our Python scripts from thousands of lines to nine, and even more exciting features are coming soon thanks to Pandorabots technology.”

- Jens Edringer

I’m using Pandorabots to power Back Talk and Back Talk 2; a profitable, top selling mobile entertainment application with over a million downloads. The app features a conversational, fully interactive 3D character named Izar. Pandorabots free DIY tools and services made it extremely easy to develop Izar’s artificial intelligence and chatbot conversational abilities with a robust knowledge base of facts, trivia, and jokes.”

- Brian Rigsby

I used Pandorabots to power Skyvi, a.k.a. “Siri for Android:” a mobile application that used voice recognition and speech synthesis technology to enable people to talk to their very own personal assistant in natural language. Skyvi obeyed commands like look up the weather or post to Facebook, and even held witty, intelligent conversations. With close to ten million downloads, our users loved the app and rated Skyvi 4.5 stars on the Play Store!”

- Jeff Chen

I’m using Pandorabots to power the George Artificial Intelligence Engine 2.0 for, a conversational chatbot application with a knowledge base that draws on closed-captioned text from major news television programs. The ability to generate my own text-simplified AIML patterns is especially useful because it allows me to perform better pattern matching against a data set that includes over 14,000 recorded television shows.”

- William Johnston