7.6 Progression Seeding

Progression Seeding lets your school import existing student evidence into CurricuLLM's learning profile system, so that when students start using the platform, the AI already has historical data about their ability levels.

Why it matters

Without seeding, CurricuLLM starts with an empty picture of each student. The platform only learns about students from their in-app interactions — like learning check-ins and conversations. This means it takes time for the AI to build up enough data to personalise effectively.

With seeding, you can import years of reports, work samples, and teacher observations before students even log in. The AI then has a baseline understanding of each student's strengths and gaps from day one, making personalisation immediate rather than something that builds up over weeks of use.

Where to find it

Navigate to Settings > Progression Seeding. Only administrators with organisation management permissions can access this page.

Enabling seeding

The page starts with a toggle switch: Enable Progression Seeding. When off, the entire feature is dormant — no uploads are accepted and no API keys can be used. Turning it on reveals the rest of the settings.

Two ways to upload evidence

1. Manual upload (via the settings page)

For one-off uploads, fill out the form directly in the settings page:

• Student — A searchable dropdown listing all students in your organisation. Shows name and email.
• Content Type — What kind of evidence this is:
  • Teacher Report (formal assessment report)
  • Student Work (a work sample)
  • Completed Rubric (a marked rubric or marking guide)
  • Teacher Notes (observations or anecdotal notes)
• Evidence Date — When the evidence was collected. Must be in the past and within the last 12 years.
• Document — The actual file. Accepted formats: PDF, DOCX, PPTX, TXT, HTML, CSV, RTF, ODT.

Click Upload Document to process the file and queue it for analysis.

2. API upload (for bulk imports)

For schools that need to upload hundreds or thousands of documents, the REST API allows programmatic uploads. This is designed for SIS vendors, school IT teams, or batch scripts.

The settings page includes ready-to-use Python and Node.js code examples showing exactly how to call the API, making it straightforward for school IT staff to integrate.

API key management

To authenticate API uploads, generate API keys from the settings page:

• Click Generate Key with an optional description
• The full key is shown once with a copy button — it is never shown again (only a secure hash is stored)
• Keys can be disabled (temporarily suspended without deleting) or deleted permanently
• Each key shows its description, creation date, and active status

All key operations (create, enable, disable, delete) are logged as security audit events.

What happens when a document is uploaded

Whether uploaded manually or via API, every document goes through the same pipeline:

• Validation — Checks that the student email, content type, and date are valid
• File type check — Confirms the file is one of the supported formats
• Student lookup — Verifies the student exists in your organisation with a Student licence
• Text extraction — Extracts readable text from the document (PDF, DOCX, PPTX, and other formats are all handled)
• Duplicate detection — Checks whether the same content has already been uploaded for that student. If it has, the upload is rejected with a clear message. This prevents the same report being imported twice.
• Secure storage — The original file and extracted text are stored securely
• Analysis — The extracted text is analysed against the curriculum to identify evidence of student understanding, and progression data is written to the student's learning profile

Upload status tracking

Each upload goes through these statuses:

• Pending — Queued, waiting for analysis
• Processing — The system is analysing the document
• Completed — Evidence has been extracted and added to the student's learning profile
• Failed — Something went wrong during analysis
• Rejected — The document was rejected (e.g. no usable content, duplicate)

When uploads are in progress, the settings page shows:

• A progress bar showing "X of Y complete"
• Counts of pending, processing, failed, and rejected uploads
• A "View all" link to the full upload history

The page auto-refreshes every 5 seconds while uploads are in progress, so you can watch the progress bar update in real time.

Security

• API keys are hashed before storage — the plain key is only shown once at creation time
• Manual uploads use your normal login session with organisation-level access checks
• You can only upload documents for students in your own organisation
• Documents are stored in a dedicated secure container separate from other app storage
• All seeding actions (toggle, key management, uploads) are logged as security audit events

Tips for administrators

• Start with a small batch of uploads (e.g. 10-20 student reports) to test the workflow before doing a full bulk import.
• Give your API keys descriptive names (e.g. "Compass export script" or "2025 semester reports") so you can track which key was used for which batch.
• Check the upload history regularly during bulk imports — if you see failures or rejections, address them early rather than waiting until the end.
• Remember that duplicate detection prevents the same document being imported twice, so it is safe to re-run a batch script if it was interrupted.

Summary

Progression Seeding lets your school fast-track personalisation by importing existing student evidence. Instead of starting from zero, CurricuLLM begins with a picture of each student's strengths and gaps built from their actual reports and work samples. This means better-tailored support from the very first conversation.