Marking Automation: What is it?
The Marking Automation System
is an AI-powered solution designed to automate the entire student marking process for coaching companies. Traditionally, marking student papers could take hours or even days, consuming valuable staff resources and costing the company hundreds of thousands of dollars each year.
This system uses Optical Character Recognition (OCR) technology and AI to mark both short written answers and multiple-choice questions with high accuracy and consistency. The AI is trained on a backlog of marking data, ensuring that every student’s work is evaluated fairly and consistently. Once marked, the results are instantly uploaded for student access.
While the marking is fully automated, the system allows for manual feedback if students need clarification. By eliminating the manual marking process, this automation has freed up staff time, reduced operational costs, and improved the accuracy and speed of the marking process, enabling the company to focus on business development and expanding their course offerings.
The Problem: Manual Marking Draining Resources
Inefficiency and Time Loss: The coaching company previously relied on manual marking for every student’s paper, which could take several staff members hours or even days to complete. This inefficiency detracted from business development efforts and was a drain on valuable time.
High Operational Costs: Maintaining this manual process required significant staffing resources, costing the company hundreds of thousands of dollars annually when it was not necessary. The labour costs involved in marking were unsustainable for a growing business.
Inconsistent Marking: Due to the variability in manual marking, there were frequent inconsistencies in how student answers were evaluated. This impacted the reliability of results, requiring staff to revisit papers and make adjustments.
The Solution: OCR Automation
To tackle the time and cost inefficiencies of manual marking, we developed a fully automated marking system tailored to the company's needs. The solution integrates the following key features:
AI-Driven OCR for Marking: The core of the system is an Optical Character Recognition (OCR) AI that was trained on an extensive backlog of marking data. This AI is capable of recognising, interpreting, and evaluating both short written responses and multiple-choice answers with a high degree of accuracy. By systematically referencing the information stored in a centralised Airtable database, the AI ensures that the marking process is consistent and unbiased across all students.
Instant Results Upload: Once the AI marks a paper, the results are immediately uploaded for student access. This drastically reduces the time between submission and feedback, enhancing the student experience and creating greater transparency in the process.
Complex Workflows: Custom workflows were developed for handling various types of assessments, including both short-form answers and multiple-choice questions. These workflows allow for efficient processing of large volumes of submissions without any manual intervention.
Manual Feedback for Clarification: While the marking is fully automated, the system allows for manual feedback in case students request clarification on their results. This ensures a balance between automation and personalisation, providing students with the support they need while keeping the system efficient.
Database Built on Airtable: The centralised database, built using Airtable, stores all student records, marking criteria, and historical data. This ensures that all grading is referenced and documented, allowing for future improvements to the system’s accuracy and consistency.
No More Manual Work: The manual marking process that previously required several staff members has been entirely eliminated. These staff are now free to focus on business development, expanding the company’s course offerings and growing their operations.
Deliverables
Cost savings: The system is projected to save $300,000 annually, eliminating the need for multiple staff members who were previously dedicated to the manual marking process.
Time savings: The automation saves around 3,000 staff hours annually, previously spent on manually marking and inputting results.
Improved efficiency: The system reduces marking time from days to just minutes, instantly uploading results and freeing up staff for business development.
OpenAI: OCR from Artificially intelligent language models were trained to recongise every single pilots unique handwriting, Significant effort was taken to ensure it is accurate 99% of the time.
Airtable: We used Airtables features to build out a company data-base that our AI language would automatically insert and organise
Parser: Using a parser we broke down every single piece of data collected from OCR and made sure it would end up in the correct corresponding location in the new data-base.
Having data broken down in such a way allowed another custom trained AI language model use data from OCR, and build out a custom invoice for specific clients.
Integrated Technology: Automations we created were integrated with the companies existing technology stack only having to pay subscriptions to two new softwares.
Quality Checks: Quality checks were added within the final stages of the automation for front of office staff to double check the accuracy, when done a single button is clicked to finish the workflow and generate an invoice for sending.