This website values your privacy and does not use cookies.

Grady Logo

The Dream
Teaching Assistant

We designed our dream teaching assistant. Grady grades student papers, guides students on what to improve, and prevents cheating.

Empowers Faculty with Full Control

Gives instructors complete control over the entire grading process.

Outstanding Learning Outcomes

Academic excellence at scale through the highest quality grading standards.

Ethical and Safe AI

Every component reflects our commitment to responsible AI.

Genuine Data Privacy

Embedded into the system's architecture as a foundational principle.

Team

Founders

Periklis Papakonstantinou

Periklis A. Papakonstantinou

Co-Founder & CEO

Associate Professor, Rutgers University, The State University of New Jersey. Researcher in AI and cryptography. Funded by NSF & NSF-China.

Anastasios Sidiropoulos

Anastasios Sidiropoulos

Co-Founder & CTO

Associate Professor, University of Illinois at Chicago. Researcher in AI and algorithms. Secured $1M+ in NSF funding.

◆

Our founders have taught and conducted research at MIT, Tsinghua University, University of Toronto, Columbia, NYU, University of Illinois UIC & UIUC, Rutgers University, Toyota Technological Institute at Chicago, The Ohio State University, and Google Research.

Corporate Advisory Board

Anant Agarwal

Anant Agarwal

Strategic Growth Advisor

Chief Academic Officer at 2U
Founder and CEO of edX
Professor
MIT

Nina Huntemann

Nina Huntemann

Strategic Growth Advisor

EdTech executive
Former VP of Learning at edX
Chief Academic Officer at Chegg

Mark Haseltine

Mark Haseltine

Strategic Growth Advisor

Product, AI & Engineering
Head of Product, Manifest
Former CEO, Teachable
Former CTO, edX

Brett Chereskin

Brett Chereskin

Advisor

COO, dub
Fintech Executive, Veteran

Academic Relations Advisory Board

Jaideep Vaidya

Jaideep Vaidya

Academic Relations Advisor

Vice Dean and
Distinguished Professor
Rutgers University
Vice-President publications (2024), IEEE-CS

Yaw Mensah

Yaw Mensah

Academic Relations Advisor

Vice Dean and Professor
Rutgers University

Roman Holowinsky

Roman Holowinsky

Academic Relations Advisor

Associate Professor
The Ohio State University
Founder, Erdős Institute
Co-Founder, STEAM Factory

Technical Discipline Advisory Board

Alexander Paulin

Alexander Paulin

Discipline Advisor

Associate Teaching Professor of Mathematics
UC Berkeley

Austin Smith

Austin Smith

Discipline Advisor

Lecturer of Creative Writing
Stanford University
Poet and Fiction Writer

Nakul Verma

Nakul Verma

Discipline Advisor

Senior Teaching Faculty of Computer Science
Columbia University
Director of the Data Science Master program

Tim Carpenter

Tim Carpenter

Discipline Advisor

Senior Lecturer, Computer Science and Engineering
The Ohio State University
Veteran

Peggy Heffington

Peggy Heffington

Discipline Advisor

Professor of History
University of Chicago
Writer

Oscar Pineda-Catalan

Oscar Pineda-Catalan

Discipline Advisor

Associate Senior Instructional Professor of Biological Sciences
University of Chicago

What is Grady

Grady is the only autonomous grading system available today. Its ecosystem goes beyond grading, featuring the (1) Instructor Course Summarization Dashboard; (2) Student Guidance and Progress Dashboard; (3) Proprietary Academic Integrity Engine.

The Autonomous Grading Model

Empowers Educators

Instructors provide high-level guidance, retaining full control grading.

Intelligent & Scalable

Grady generalizes grading instructions and applies them consistently across any number of submissions.

Instructor-Tuned Grading

Instructors can choose to guide grading at any level of detail: from high-level remarks to a full rubric, reference answer, and regrading tweaks.

Superhuman Accuracy

Achieves superior grading accuracy, validated in controlled human trials with 99% statistical confidence.

Research, Development, and IP

Foundational Research

Designed and developed by leading researchers with expertise in Machine Learning, Data Privacy, and Information Security.

Rigorous Validation

Grady exhibits superhuman grading accuracy in randomized controlled human trials.

Patents (US patent pending)

AI-Based System and Method for Detecting Academic Dishonesty by Analyzing Concept Mastery Over Time

A System and Method for AI Driven Contextual Rubric Generator with Adjustable Leniency Control

Our Ethical & Responsible AI Pledge

Grady is built not only on unique R&D but foundationally based on responsible innovation.

Responsible Use of AI

Our algorithms are continuously studied for transparency, bias, and fairness because we are eduators and serving education is our misssion.

Genuine Data Privacy

Designed from the ground up on principles of Differential Privacy, our proprietary anonymization engine ensures genuine data protection beyond boilerplate compliance.

Academic Integrity

We innovate in cheating prevention and actively safeguarding academic integrity.

Student & Learning Outcomes

Instructors receive cumulative class resports and students receive personalized reports that identify actionable items for improved learning outcomes.

Documents and Whitepapers

We believe in transparency and rigorous documentation. Key resources are available to our current and prospective institutional partners upon request.

Data Privacy Policy

Our policy details our proprietary, differentially private anonymization engine—a standard of security no competitor can match.

Accessibility (VPAT/WCAG 2.1)

Our Voluntary Product Accessibility Template (VPAT) outlines conformance with accessibility standards.

Technical Whitepapers

Detailed reports on our algorithms, human trials, and security architecture are available upon request.

Request Access →

Contact Us

We welcome inquiries from university administrators, department chairs, and program committees. To learn more or to schedule a personalized demonstration, reach us at: corporate@gradyai.com

Request a Demo →

Determinant Technologies Privacy Policy

Effective Date: July 17, 2025

This Privacy Policy describes how Determinant Technologies and our autonomous grading service, Grady, collect, use, and protect the information we process when you use our services. Our commitment is rooted in a "privacy-by-design" approach, ensuring that student and institutional data is handled with the highest standard of care.

This public-facing policy provides a transparent overview of our practices. Determinant Technologies also maintains a complete and detailed internal Data Privacy policy (TR-2025-004), which is available to our institutional partners under a Non-Disclosure Agreement (NDA).

Our Role: Data Processor

It is important to understand the roles regarding your data.

  • • The Data Controller is the educational institution (e.g., your university, school, or online educational platform or program). They own the student data and determine the purposes for its use.
  • • The Data Processor is Determinant Technologies. We process data strictly on behalf of and under the instruction of the institution for the sole purpose of providing our grading service, Grady.

All student requests regarding their data, such as access and deletion, should be directed to their institution's administrators. We are fully committed to assisting our institutional partners in fulfilling these requests.

What Information We Process (and What We Don't)

Our system is engineered on the principle of data minimization. We only process the information absolutely necessary to perform our grading service.

We Process: The content of a student's submission for a specific assignment. This is sent to us along with internal, non-identifying IDs from the Learning Management System (LMS) (e.g., Canvas) so we can return the grade to the correct place.

We Do NOT Process: We never ingest, process, or store any Personally Identifiable Information (PII). This includes, but is not limited to:

  • Student or instructor names
  • Email addresses
  • Student ID numbers
  • IP addresses or device information (server logs are purged in near real-time)

How We Process and Protect Information

Our entire data flow is designed for maximum security and privacy.

  • Secure Transfer: An instructor initiates grading through the LMS. The assignment content is sent to Grady over a secure, encrypted connection.
  • Immediate Anonymization: The moment we receive a submission, it is processed by our proprietary anonymization technology. This advanced process sanitizes, paraphrases, and transforms the text to retain academic meaning while making it impossible to trace back to the original author. The original submission is securely destroyed as soon as the data are sanitized, which is typically within two minutes.
  • Secure, Ephemeral Grading: The now fully anonymized content is graded within a secure, confidential computing environment (such as AWS Nitro Enclaves).
  • Grade Return: The score and feedback are returned to the institution's LMS.
  • Zero Retention: Once the grade is returned, the anonymized version of the submission is permanently and securely wiped from our systems, typically within 30 minutes. Nothing persists on our servers.

Our Core Privacy Principles

Our commitment to you is defined by these core principles:

  • Data Minimization: We only ever receive the assignment content itself, never personal identifiers.
  • Ephemeral Processing: Data exists on our systems only for the few minutes needed to grade it. We operate on a strict zero-retention policy post-grading.
  • Deep Anonymization: We use a proprietary process to paraphrase and de-identify all submissions before analysis. This ensures that the content sent to AI models cannot be linked to an individual student.
  • Synthetic Training Data Only: We do not use student data (not even anonymized versions) to train our AI models. Our models are trained exclusively on high-quality, synthetically generated data created in-house.
  • No Tracking: We do not use cookies or any other session or device tracking technologies.
  • No Data Sale: We will never sell, rent, or share student or institutional data. Our business model is based solely on providing our grading service.
  • US Data Residency: All data is processed on secure servers located within the United States.

Compliance with Data Protection Laws

Our "privacy-by-design" architecture inherently complies with major data protection regulations.

  • FERPA: We act as a "school official" under FERPA, using data only for legitimate educational interests as directed by the institution and never re-disclosing it.
  • US State Privacy Laws (CCPA/CPRA, etc.): We act as a "Service Provider." Since we do not collect personal information or retain any data, we inherently satisfy the core requirements of these laws regarding data access, deletion, and the right to opt-out of data sales.
  • GDPR: For institutions serving EU citizens, we act as a Data Processor and provide a Data Processing Agreement (DPA). We have a special GDPR Operational Mode that guarantees a human instructor has the final, absolute authority to review and override any AI-generated grade, ensuring a "right to human intervention."

Third-Party Services

We use a limited number of trusted third-party services to provide our service.

  • Cloud Infrastructure: We use Amazon Web Services (AWS) for secure compute and storage, located in the US.
  • AI Models: We send only the fully anonymized and paraphrased text to AI providers for analysis. We have contractual agreements and technical controls in place with these providers to ensure this anonymized data is not logged or used for their model training.

Data Requests and Governance

  • Student Data Requests: As we hold no personal data, requests for access or deletion are best handled by your institution. We will respond within two business days to any institutional request to formally confirm that we hold no data related to a specific user.
  • Government & Law Enforcement Requests: We will only disclose data if compelled by a valid and legally binding order. However, our zero-retention architecture is our primary safeguard; in practice, no student or institutional data exists on our servers to be produced.
  • Policy Governance: This policy is maintained and reviewed annually by our Chief Information Security & Privacy Officer (CISPO), Periklis A. Papakonstantinou. Professor Papakonstantinou is a distinguished expert in Cryptography, Cybersecurity, and Data Privacy who oversees our privacy program and ensures our practices meet the highest ethical and legal standards.

Changes to This Policy

We may update this Privacy Policy from time to time. When we do, we will post the updated version on this page and revise the "Effective Date" at the top.

Contact Us

If you are an institutional partner and have questions about this policy or our privacy practices, please contact our CISPO at corporate@gradyai.com.

Accessibility Conformance Report (VPAT)

Name of Product/Version: Grady, v0.1

Report Date: July 20, 2025

Product Description: The Grady Autonomous Grading LTI tool is a software application designed to integrate with the Instructure Canvas Learning Management System (LMS). It provides instructors with AI-powered assistance for grading assignments and quizzes. Key functionalities include automated extraction of course documents from the syllabus, management of grading status for assignments, configuration of grading parameters (such as uploading reference answers and rubrics), and a review and approval workflow that emulates the user experience of the native Canvas SpeedGrader interface. The tool relies exclusively on standard Canvas UI elements for all user interactions, including but not limited to file uploads, text boxes, buttons, and navigation sequences.

Contact Information: Determinant Techologies, Inc. corporate@gradyai.com

Notes: This Accessibility Conformance Report (ACR) provides a presumptive assessment of the Grady Autonomous Grading LTI tool's conformance with relevant accessibility standards. The tool is architected as a Learning Tools Interoperability (LTI) application that operates entirely within and utilizes the user interface (UI) components of the Instructure Canvas LMS. The conformance level assessments documented within this report are therefore based on a functional analysis of the Grady tool, cross-referenced against the publicly available Accessibility Conformance Report for the Canvas LMS (based on VPAT® Version 2.4, dated November 2023), as provided by Instructure.(https://www.instructure.com/products/canvas/accessibility)

Evaluation Methods Used: The evaluation method used to produce this report is a functional analysis of the Grady Autonomous Grading LTI tool's user interface and workflows. This analysis was cross-referenced against the accessibility claims made in the official Canvas LMS Accessibility Conformance Report (VPAT® v2.4, November 2023). The evaluation included manual testing with screen readers (NVDA for Windows and Apple VoiceOver for macOS), keyboard-only navigation testing, and automated accessibility checkers. The stated conformance levels are based on this testing and the fact that Grady relies on native Canvas UI elements and thus behaves identically to the underlying platform from an accessibility perspective.

Applicable Standards/Guidelines

Standard/GuidelineIncluded In Report
Web Content Accessibility Guidelines 2.0Level A (Yes) YES,Level AA (Yes) YES,Level AAA (Yes / No) NO
"Revised Section 508 standards published January 18, 2017 and corrected January 22, 2018 "(Yes) YES

WCAG 2.0 Report

Table 1: Success Criteria, Level A

CriteriaConformance LevelRemarks and Explanations
1.1.1 Non-text Content (Level A) Web: Supports Web: Functionality: The Grady tool's interface includes non-text content in the form of interactive controls, such as buttons ("X" to remove a file, file upload buttons, "Start Grading," "View Rubric," "Regrade everything," etc.). Platform Reliance: These controls are standard Canvas UI elements. The Canvas LMS ACR reports Supports for this criterion, stating that it provides text alternatives for default non-text content.
1.3.1 Info and Relationships (Level A) Web: Partially Supports Web: The Grady tool presents information in structured ways, such as lists of course documents and assignments, and uses form elements for user input. The review interface is described as being similar to Canvas SpeedGrader, which relies on a specific layout structure. Canvas, within which Grady is running and uses its native controls also reports “Partially Supports”
2.1.1 Keyboard (Level A) Web: Partially Supports Web: All interactive elements within the Grady tool, including buttons, links, form fields, and navigation tabs, are intended to be fully operable via a keyboard. The user workflow involves navigating lists, activating buttons, and entering text into fields. The exception is the navigation buttons as in the Canvas Speedgrader. Canvas reports here “Supports”, but we could not verify this for Speedgrader.
2.1.2 No Keyboard Trap (Level A) Web: Supports Web: Users can navigate to and from all interactive elements using only the keyboard without their focus becoming trapped within a component.
2.2.1 Timing Adjustable (Level A) Web: Supports Web: The Grady tool does not impose any time limits on user interaction. The grading process itself is asynchronous and can take from 30 minutes to several hours, but this is a background process and does not time out the user's session. The user is notified by email upon completion.
2.4.2 Page Titled (Level A) Web: Supports Web: The Grady tool presents several different views or pages to the user (e.g., main dashboard, assignment configuration screen, grading review screen).
4.1.1 Parsing (Level A) Supports For WCAG 2.0 and the 508 standards, the September 2023 errata update indicates this criterion is always supported. See the WCAG 2.0 Editorial Errata.

Table 2: Success Criteria, Level AA

CriteriaConformance LevelRemarks and Explanations
1.4.3 Contrast (Minimum) (Level AA) Web: Supports Web: The visual presentation of all text and UI components in the Grady tool is rendered using the styling of the host Canvas LMS.
1.4.4 Resize text (Level AA) Web: Supports Web: The Grady tool's interface is usable without requiring two-dimensional scrolling when the browser is zoomed to 400% (equivalent to a 320 CSS pixel width viewport).
2.4.7 Focus Visible (Level AA) Web: Supports Web: When a user navigates the Grady tool's interface using a keyboard, the component that currently has focus must have a highly visible focus indicator.
3.2.4 Consistent Identification (Level AA) Web: Supports Web: Same UI items (such as press “X” to remove a file) are consistently used.

Revised Section 508 Report

Chapter 3: Functional Performance Criteria (FPC)

CriteriaConformance LevelRemarks and Explanations
302.1 Without VisionSupportsThe product is a software application that can be used by individuals who are blind. By leveraging the Canvas LMS platform, which is optimized for screen readers, the Grady tool is operable without vision. This has been verified through testing with NVDA (for Windows) and Apple VoiceOver (for macOS) screen readers.
302.2 With Limited VisionSupportsSee 302.1
302.3 Without Perception of ColorSupportsOne of the designers of Grady, who also tests the product, is colorblind.

Chapter 5: Software

CriteriaConformance LevelRemarks and Explanations
501.1 Scope – Incorporation of WCAG 2.0 AASee WCAG 2.x sectionSee information in WCAG 2.x section
502.2.1 User Control of Accessibility FeaturesSupportsThe software is designed to interoperate with assistive technology by virtue of its integration with the Canvas LMS platform. Conformance has been verified through testing, confirming that the tool does not disrupt and allows user control of accessibility features.
502.2.2 No Disruption of Accessibility FeaturesSupportsSee above.
503.2 User PreferencesSupportsThe product is a web-based LTI application and does not override user-selected preferences from the operating system or browser, such as color, contrast, and font type. These settings are passed through and rendered by the browser and the host Canvas platform.

Chapter 6: Support Documentation and Services

CriteriaConformance LevelRemarks and Explanations
602.2 Accessibility and Compatibility FeaturesSupportsWe provide electronic support documentation and a 3-minute explanatory video. The documentation is in PDF form and thus comes with the WCGA accessibility features of standard PDF readers, such as Adobe Acrobat Reader. The video has synchronized captions English captions.
602.4 Alternate Formats for Non-Electronic Support DocumentationSupportsDeterminant Technologies can mail a printout of the documentation in larger print as requested.