Skip to content

Home

Jump to bottom
Rafał edited this page May 27, 2023 · 2 revisions

Links:

Intro

During LabLab AI Anthropic Hackathon we plan to focus on leveraging Anthropic AI models to bring a transformative feature to SmartSnap application - The "Text Analyzer".

The SmartSnap app, already serving close to 1,000 users in Poland, is aimed at making learning more accessible, interactive, and engaging for students. It uses AI to assist users in solving problems across various academic subjects by merely snapping a photo of the problem. Building on this success, we're eager to enhance our application's capabilities even further during the hackathon.

The "Text Analyzer" feature will change how students interact with texts from reading comprehension tests or English language lessons in our app. After snapping a photo or entering text, users will be able to ask for a summary or pose text-based exercises. The model will provide the user with detailed answers and comprehensive explanations, serving as a personal learning assistant to facilitate a deeper understanding of the content. For example, the AI could analyze a passage and answer true/false questions about the text.

Development details

  • We will create language-pal repo that will be based on our new math-pal repo for v4 math API endpoint.
  • We will use langchain and Claude API

Endpoints:

Function name: v4_analyzeText

Description:

This function prepares Claude AI model to work on given text.

Endpoint:

POST /v4_analyzeText

Request body:

  • text: A string containing the text for analysis. This is required.
  • language: A string indicating the language of the text. Default is "en". This is optional.

Response:

{
  "context": string
}
  • context: A string representing context that will be later used to ask question about the given text

Function name: v4_summarizeText

Description:

This function returns summary for a given text.

Endpoint:

POST /v4_summarizeText

Request body:

  • context: A string representing context that will be later used to ask question about the given text.

Response:

{
  "summary": string
  "context": string
}
  • summary: A string containing the summary text
  • context: A string representing context that will be used to ask another question about the given text

Function name: v4_askQuestionText

Description:

This function ask a question for the given text.

Endpoint:

POST /v4_askQuestionText

Request body:

  • context: A string representing context that will be used to ask question about the given text
  • problem: A string containing problem that we want to solve for this text. This is required.
  • language: A string indicating the language of the solution. Default is "en". This is optional.

Response:

{
  "question": string
  "solution": string
  "context": string
}
  • question: A string containing the problem that we wanted to solve for this text
  • solution: A string containing solution for given problem.
  • context: A string representing context that will be used to ask question about the given text

Clone this wiki locally