The Software
Version 2.5 (July 2026)
How it works
Add your PDFs, articles, or text documents. The software automatically detects and organizes them.
The software applies the same criteria to each document individually — data is automatically tabulated.
Results are consolidated into an exportable report with syntheses, clusters, and full traceability.
Time
It depends on the number of data points to be analyzed. Our average estimate is approximately 1 hour per 100 documents.
Performance
According to our tests, data extraction achieves near 100% accuracy — surpassing manual reading in exhaustiveness and consistency. Errors are still possible, which is why every analysis is traceable and reviewable. Reliability is further reinforced by adequacy crosschecking analyses, which verify the correspondence between what was extracted and the original documents.
Reliability
FullCorpus AI is not an AI or an LLM. It is software that automates the use of LLMs (GPT, Gemini, and Claude). Its reliability is therefore on par with high-effort, high-reasoning models.
Our differentiator
Data is not mixed in bulk — each document is analyzed individually, with the same criteria applied one by one. It is like running the same prompt for every document, on a model with the highest reasoning effort.
Application
We recommend it as a pre-analysis stage — an initial round of analysis, often also simulating possible results and findings. Researcher review and analysis are still necessary.
We use developer APIs from GPT, Gemini, and Claude — with greater context capacity and performance than chat interfaces.
While chatbots answer one question at a time, FullCorpus AI applies the same criteria to hundreds of documents with fidelity and consistency.
Errors can occur. That is why every analysis is documented and reviewable — traceability is part of the process, not an add-on.
Features
Analyze dozens or hundreds of documents in sequence, with the same prompt and criteria applied to each one.
Choose GPT, Gemini, Claude, or others for each task. Switch models without reconfiguring the workflow.
Results in tabular format (.xlsx, .csv) with one row per document, ready for statistical analysis.
Every analysis records the model, version, temperature, and prompt used — with a unique identification hash.
Import PDFs, scientific articles, .txt and .docx documents. The software prepares and normalizes them automatically.
Analyze documents in Portuguese, English, Spanish, and other languages — with prompts in any language.
Compatibility
Early access
We are selecting researchers and teams for our early access program. Get in touch and tell us about your corpus.
Academic reference
If you use FullCorpus AI in your research, please cite it as [APA 7th ed.]:
