Case study · Education
Faster feedback, still written by teachers
Teachers spend evenings writing feedback on assignments. This is how we built an assistant that drafts feedback against the teacher's own rubric, which the teacher edits and approves, with student data protected and age-appropriate safeguards in place.
Project overview
Teacher-reviewed AI feedback.
- Industry
- Education
- Client
- Confidential
- Solution
- Feedback assistant
- Platform
- Azure OpenAI + LMS
- Timeline
- 10–14 weeks
- Team
- 3 people
- Input
- Assignments and rubrics
- Review
- Teacher approves every comment
- Data
- Student data protected
Expected business outcomes
What the client set out to achieve
Measured with teachers in a pilot.
-
30–50%
Less time writing feedback
Teachers edit rubric-aligned drafts.
Target range; measured in a timed pilot
-
100%
Feedback approved by a teacher
No AI comment reaches students without review.
Enforced by workflow
-
Faster
Feedback to students
Students receive feedback sooner.
Measured as days from submission
Business challenges
Where the client started
Good feedback takes time teachers don't have.
Workload
Marking and feedback happen in evenings and weekends.
Timing
Students wait a long time for feedback.
Consistency
Feedback varies between teachers.
Rubrics
Rubrics exist but take time to apply to each piece.
Concern
Leaders worry about AI, academic integrity and child safety.
Proposed solution
What the assistant covers
Drafting against the teacher's rubric, never automatic grading.
-
Rubric-based drafts
Feedback drafted against each rubric criterion.
-
Teacher review
Teachers edit, approve or reject every comment.
-
Safeguards
Content filters and escalation for safeguarding concerns.
-
Student data protection
Processing in a secure tenant, no data used for training.
-
LMS integration
Assignments read from and feedback returned to the LMS.
-
Quality monitoring
Teacher edits and ratings reviewed by the teaching lead.
Project requirements
What the client asked us for
The request was to reduce workload without reducing quality. The assistant drafts feedback; teachers remain the authors.
Safeguarding and data protection assessments come first, with age-appropriate controls.
What the project had to deliver
- 01Rubric-aligned drafts
- 02Teacher approval always
- 03Safeguarding built in
- 04Student data protected
Key challenges & considerations
What made it harder, and what we did
-
Safeguarding obligations
What we did
Safeguarding reviews were completed before any student work was processed.
-
Varied subjects and rubrics
What we did
Rubrics were configured per subject with the teachers.
-
LMS integration limits
What we did
The assistant worked within what the learning platform allowed, with export where it did not.
-
Teacher trust in AI
What we did
Teachers stayed the authors: every comment was editable, and volunteers went first.
What we watched for
- Complete safeguarding reviews first
- Keep teachers as authors
- Start with volunteer teachers
- Monitor edits for quality
Development process
Safeguards, pilot, scale
Governance comes before any student work is processed.
-
Governance
Safeguarding and data protection reviews.
-
Rubrics
Rubrics digitised with teaching leads.
-
Build
Drafting, review and LMS integration.
-
Pilot
Pilot with volunteer teachers in selected subjects.
-
Scale
Expanded by subject with monitoring.
Technology stack
Responsible AI for education
Private models and strict safeguards.
AI
- Azure OpenAI
- Retrieval-augmented generation
- Prompt and output guardrails
Data
- Azure AI Search
- Document pipelines
- Vector index
Application
- Python / .NET services
- Web or Teams interface
Evaluation
- Test question sets
- Quality and safety metrics
- Usage monitoring
Quality assurance & deployment
Measured against a test set before launch
AI quality was measured, not judged from a demo. A test set of real examples with agreed answers decided whether the system was ready, and the same set was re-run after every change.
- A test set of real examples with agreed answers
- Accuracy thresholds agreed with the business before launch
- Safety checks: refusals, permissions and prompt-injection tests
- A person reviewing low-confidence output
- Cost and response time measured at real volumes
Deployment
- 01PilotPilot with volunteer teachers in selected subjects.
- 02ScaleExpanded by subject with monitoring.
- 03Pilot, then staged rolloutA small group of real users first, with results reviewed weekly before widening access.
- 04Monitoring and warrantyAccuracy, drift, usage and cost tracked after launch, and six months of free bug fixing.
Final deliverables
Feedback sooner, teachers in charge
Teachers start from rubric-aligned drafts and finish feedback in less time, and students receive it sooner.
Every comment is still approved by a teacher, and safeguards are in place.
- Rubric-aligned drafts
- Teacher approval
- Safeguards
- Faster feedback
Handed over to the client
- 01Production AI serviceDeployed in the client's own cloud tenant
- 02Source code and modelsIn a repository the client can access
- 03Test set and resultsThe evidence behind the launch decision
- 04MonitoringAccuracy, usage and cost
- 05Six-month warrantyFree bug fixing after delivery
Success metrics
How success is measured
| Measure | Outcome | How it is measured |
|---|---|---|
| Less time writing feedback | 30–50% | Target range; measured in a timed pilot |
| Feedback approved by a teacher | 100% | Enforced by workflow |
| Feedback to students | Faster | Measured as days from submission |
Verified reviews
What clients say about our custom software work
Verified reviews from our clients on the same kind of work, published on Clutch.
-
Custom Software Development
Web-Based Coupon App Development
The work produced by the team was of exceptional quality.
Jon Thies
CTO, Model Rocket Feb 2022 – Mar 2023
-
Custom Software Development
Mobile & Web Platform Development & Design
Silver Scintilla was responsive, respectful, and enjoyable to work with, quickly addressing any questions or requests.
DeShawn Brown
CEO & Founder, Lithios Jun 2021 – Nov 2022
-
Custom Software Development
Web-Based HRM System Development
Silver Scintilla provided the best solution with scalable functionality, personalized features, and robust security.
Suheb Khan
Director, Collaborative Insight Technologies Jul 2020 – Jun 2022
Free discovery call
Have an idea? Let's turn it into AI-powered software.
Book a free discovery call with our experts. Share your idea and we will help you shape the scope, timeline and budget, under NDA.
- Free consultation
- NDA before we talk
- Transparent estimate