AlliedOffsets is excited to launch the AI Document Reader (AIDER), a tool to extract data from static project documents across the voluntary carbon market.
The AIDER tool allows users to query thousands of documents, furthering the company’s mission to promote transparency in the VCM.
AIDER was developed for in-house use – and now we are opening up the tool for external clients and stakeholders. It complements our existing offerings, including the Advanced Document Search, VCM Alerts, and the VCM Price Calculator.
AlliedOffsets has used AIDER to extract a number of highly valuable data points, which we surface to clients of our dashboard, including:
- fNRB values for cookstove projects
- Plant and animal species protected by NBS projects
- Biodiversity metrics
- Estimated project emission reduction forecasting
- Project proponent information
- Self-assessed risk values
- And much more
AIDER uses OpenAI’s ChatGPT technology to loop a single prompt through documents that have been uploaded to our proprietary document database. Currently, there are over 2,500 documents across five registries that are available for querying – with more documents being added daily.
The ideal use case is for specialized data extraction. Instead of manually searching through hundreds of documents (PDDs, verification reports, monitoring reports, and others) to pick out valuable data, AIDER uses the latest AI technology to extract the data in a fast, scalable, and accurate manner.
The example below highlights how AIDER can be used to extract data on species protected by projects in the VCM.
First, users select the project documents they are interested in. They can filter by a number of parameters, including registry, sector, and document type.
Next, they create the prompt they would like to loop through the selected documents. Users can be as specific as they’d like.
Finally, the data comes back for each of the documents queried.
User can download the results as a .csv file in order to incorporate the data into their data processes or applications.
Over the coming weeks, we’ll share results of some of the queries we’ve found helpful ourselves – but the possibilities are limitless. If you’d like to learn more about the tool and how it can help you make sense of the VCM, please get in touch.
