Brainly
Snap to Solve photo search for homework questions
Brainly's photo search drew 6x more engagement than typed questions, and 70% of users were happy with results
Problem
Students use phones for homework help, and typing questions on a smartphone is cumbersome. Brainly runs in 35 countries, so any new feature had to work in many languages.
Approach
Brainly built Snap to Solve in its mobile app. A student photographs a question, Google's Vision AI reads the text with optical character recognition, and Brainly returns an instant answer if there is a perfect match. If not, the app falls back to a full text search of its knowledge base. Brainly's team also built an in-house model with TensorFlow's Multilingual Universal Sentence Encoder to find instant answers.
Results
Takeaway
Brainly said users who received instant answers were more likely to buy its subscription. The company expects photo search to become the most popular way to search, with voice search also growing.
Categories
- Industry
- Education
- Business function
- Marketing and sales
- Innovation type
- Product and service innovation
- AI technology
- Computer vision
- Adoption
- Off-the-shelf tool or platform
- Evidence
- Vendor case study
Sources
Source summaries
Google Cloud's customer story describes how the Polish learning platform Brainly built its Snap to Solve photo search with Vision AI and TensorFlow. It reports 6x more engagement than typed queries, ten times more instant answers than at launch and more than 70% user satisfaction. The page carries no date; the Internet Archive first saved it on March 18, 2022.
Full details are in the original sources.