Now in Limited Early Access

Vandalizer logo

Next generation document handling, built at the University of Idaho.

SIGN IN
Atomic Units of Work

Verified Tasks &
Reliable Extractions

Forget generic "chat with PDF". Vandalizer uses specialized tasks to perform specific actions with high accuracy.

Structured Extraction

Pull dates, budgets, and requirements from messy PDFs into clean JSON.

Smart Summarization

Generate executive summaries tailored to specific grant guidelines.

Extracion Task COMPLETED
"action": "extract_deadline"
"result": "extraction.csv"
Verified Prompt PROCESSING
"action": "check_compliance"
Analyzing...
End-to-End Automation

Chain Tasks into
Powerful Workflows

Combine tasks to handle complex processes. Output from one task becomes input for the next, creating reliable pipelines that scale.

  • Standardize processes across your team
  • Handle conditional logic and branching
  • Reproducible results, every time
Input: RFP.pdf
Task: Extract Requirements
Task: Draft Outline

Tools to Augment your Team

Collaborate, Verify, and Scale with confidence using our secure platform.

01

Unified Team Workspace

All your team's files and workflows in one central hub. Share specific tasks or entire pipelines instantly. No more emailing refined prompts or Python scripts back and forth.

02

Private & Secure

Powerful AI models that are private and secure. Your data stays within your control and is never used to train public models.

03

Reproducible

Eliminate variability. Build workflows that produce consistent, standardized outputs every time, regardless of who runs them.

04

Evaluate Once, Run Everywhere

Create a "Golden Set" of documents to verify a workflow's accuracy. Once a workflow passes evaluation, deploy it to your entire team with the confidence that it performs correctly.

Not a replacement.
An augmentation.

1

Upload Document

Drag and drop RFPs, contracts, or emails into the secure workspace.

2

Select Tasks

Choose from pre-built "Verified" workflows or create custom extraction logic.

3

Review & Export

Validate the AI output against the source document and export to your system.

"Vandalizer allows RAs to leverage their institutional knowledge to create flexible workflows targeted at common issues."
Input Source
Research Administrator
PDF/Email
Vandalizer Core
AI Processing & Extraction
>> Extracting key_dates...
>> Validating budget_caps...
✓ Confidence: 98.5%
Final Validation
Human Expert Approval
Verified

Born at the University of Idaho

Vandalizer is an open-source initiative developed by the Artificial Intelligence for Research Administration (AI4RA) team at the University of Idaho.

This project is made possible through the support of the NSF GRANTED program (Award #2427549), dedicated to reducing barriers to research administration and building capacity at emerging research institutions.

University of Idaho
NSF GRANTED

Help Shape the Future of RA

We are actively seeking collaborators, testers, and contributors. If you are a Research Administrator interested in piloting Vandalizer, or a developer wanting to contribute to the codebase, we want to hear from you.

Contact the Team Visit AI4RA Website