Origin, State & Demand and Future Prospects
Common bot architecture
Conversational Flow Design
Bot development and implementation
Deploying to messenger, testing and training
Metrics and Analytics
Going Live
Prospects 2. Common bot architecture 3. Conversational Flow Design 4. Bot development and implementation 5. Deploying to messenger, testing and training 6. Metrics and Analytics 7. Going Live 8. Closing
to simulate human conversations especially over the internet • Smartphones disrupted the web and led to responsive design • Responsive design birthed mobile apps with provided FUX (Frictionless User Experience) • Humans prefer to converse. 78% of users use 3 apps or less, with messaging apps topping the list. This discovery led to explosion of chatbots which is where we are today
future is artificial intelligence and data both of which are dependant on each other e.g IoT, Intelligent Business using ML, Conversational Bots etc • For brands, the future is becoming more customer centric. It lies in leveraging technology in gaining more customer insights, personalizing your brand, creating more user engagement and improving on the user experience
use case i.e Entertainment bot or business bot • Generative models generate responses from scratch. Primarily based on deep learning models and rely on the ability of the bot growing smarter. • Retrieval-Based models generate responses from repository of predefined responses. Can also incorporate ML to understand different contexts and is the one employed in many scenarios
paper • Use mockup tools such as Botsociety, Botmockup, Botframe etc to implement initial design • As a first approach, simplify the conversation and have a direct approach to it • Send it to users and iterate scenarios based on feedback • Keep the conversation simple, fluid and natural
need for flexibility and the channel to be used in selecting development platform • Converse.ai, chatfuel, octane, many chat etc common bot development platforms • Coding from scratch using languages e.g Python or C and libraries e.g Chatscript, AIML, PyBrain etc • It’s paramount to have an NLP connected to the bot (Nobody likes a stupid bot) or to build one yourself
and structure chosen • Some platforms have inbuilt analytics system • Facebook Graph API allows you to extract user information. Using this and logs, create a custom analytics dashboard linked to your DB(SQL,Firebase etc) • Use 3rd party tools like dashbot • If bot is on messenger, FB has inbuilt analytics system
machine intelligence in Africa • Currently focuses on integration into health, education and poaching • Will also encompass NLP able to understand swahili • Beta version ready • Contact: [email protected] | [email protected]