OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Search & Research AI

Autogeo

AutoGEO is an open-source Python framework for optimizing web documents for visibility in large-language-model-generated answers. It extracts content-preference rules from generative engines, then uses either a prompt-based rewriting route or a reinforcement-learning-trained model to revise documents. Its evaluation framework reports visibility and utility measures.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Autogeo — OpenViST project profile in Search & Research AI.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use AutoGEO when you want to investigate or test document revisions for generative-engine visibility against a specified dataset and engine, and can configure the required API access or training resources.

Who it is for

  • Content teams exploring how to adapt documents for generative-search answers
  • Researchers comparing original and revised documents against generative engines
  • Practitioners developing or training content-optimization models

Documented uses

  • Extract engine- and dataset-specific content preference rules
  • Rewrite a document using AutoGEO API or custom rules
  • Train an AutoGEO Mini model with supervised fine-tuning and GRPO
  • Evaluate original or rewritten documents using visibility and utility metrics
Before you choose

Requirements & limits

The project is available as a public GitHub repository under the MIT license. API keys are required for the documented API and rule-extraction setup.

  • AutoGEO API uses prompts and extracted rules; AutoGEO Mini is trained with reinforcement learning.
  • Rule extraction is tailored to the selected generative engine and dataset or domain.
  • Mini training requires two CUDA-compatible GPUs; A100 40GB or larger is recommended.
  • The documented engine families are Gemini, GPT, and Claude.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install AutoGEO for rule extraction and rewriting

Set up the documented AutoGEO API and rule-extraction environment.

Before you start

  • Access to the AutoGEO GitHub repository
  • Conda
  • API keys to configure in keys.env

Steps

  1. Clone the repository recursively with the command below.

    Documented command
    git clone --recursive https://github.com/cxcscmu/AutoGEO

  2. Change into the AutoGEO directory and run the command below.

    Documented command
    bash install.sh

  3. Activate the environment with conda activate autogeo.

  4. Configure the required API keys in keys.env.

Expected outcome

The documented environment is installed and configured for AutoGEO API use and rule extraction.

Guide limitations

  • The guide covers setup only, not successful execution.
  • Changing the generative engine or dataset/domain requires new rule extraction for AutoGEO API.
Documentation 1 ↗
Source-based practical guide

Train an AutoGEO Mini model on E-commerce data

Run the documented cold-start and GRPO training stages for an E-commerce AutoGEO Mini model.

Before you start

  • Complete the documented AutoGEO installation first
  • Two CUDA-compatible GPUs; A100 40GB or larger is recommended
  • Training data for E-commerce

Steps

  1. Activate the autogeo Conda environment.

  2. Run the command below.

    Documented command
    bash install_mini.sh

  3. Run the command below for supervised fine-tuning.

    Documented command
    bash run_cold_start.sh E-commerce

  4. Run the command below for GRPO training.

    Documented command
    bash run_grpo.sh E-commerce

Expected outcome

The documented training stages save checkpoints under `outputs/E-commerce/cold_start` and `outputs/E-commerce/grpo`.

Guide limitations

  • The project estimates about four hours for supervised fine-tuning and about 48 hours for GRPO on Researchy-GEO; those estimates are not stated for E-commerce.
  • Switching dataset/domain or engine requires retraining Mini accordingly.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
16 October 2026
Original website title
cxcscmu/AutoGEO · Official homepage title · retained in its source language
How we review sourcesSuggest a correction