Part 5 of 15 in Rules for Game-Changing Entrepreneurs← Part 4Part 6 →
This chapter explains how to put useful intelligence on familiar devices and services, including smartphones, WhatsApp, and simple Android apps. It helps builders create tools that work with intermittent connectivity, older devices, and low bandwidth.
- Put intelligence where users already are instead of requiring complex central systems.
- Examples include crop diagnosis from a phone photo and a WhatsApp AI tutor in a student’s mother tongue.
- Design for intermittent connectivity and older devices.
- Use smaller models and careful prompts so tools remain useful with low bandwidth.
- A sample prompt creates a simple Hindi WhatsApp science revision flow for Class 8 students.
Put useful intelligence where the user already is, the smartphone, WhatsApp, a simple Android app, rather than forcing users into complex central systems.
Crop-diagnosis from a phone photo or a WhatsApp AI tutor in the student’s mother tongue are edge plays. Design for intermittent connectivity and older devices. Use smaller models and careful prompts so value survives low bandwidth.
Prompt: “Create a short-response WhatsApp flow in simple Hindi for Class 8 science revision that works on basic Android phones and includes one practice question per interaction.”
Frequently asked questions
What does it mean to move intelligence to the edge?
It means putting useful intelligence on services and devices users already have, such as smartphones, WhatsApp, and simple Android apps.
What are examples of edge plays?
Crop diagnosis from a phone photo and a WhatsApp AI tutor in a student’s mother tongue are examples.
How should tools work on older devices and weak connections?
Design for intermittent connectivity and older devices. Smaller models and careful prompts can help value survive low bandwidth.
What is an example of a WhatsApp AI tutor prompt?
The chapter gives a prompt for a short-response WhatsApp flow in simple Hindi for Class 8 science revision, designed for basic Android phones with one practice question per interaction.