How TeachThemAll uses artificial intelligence

What does TeachThemAll use AI for?

TeachThemAll uses large language models (LLMs) to generate learning situations, exams, class summaries, student notes, visual resources and other pedagogical materials. The system combines the official curriculum with the teacher's methodological preferences to produce personalised content.

How it works

When you create a class, TeachThemAll sends the selected curriculum, year group and pedagogical preferences to an external AI model (currently models from OpenAI, Google and Anthropic). The model generates the content, which is automatically reviewed by quality controls before the result is shown to you. No student data is sent to the AI models.

AI models in use

We list the specific models that process your requests and what each one is used for. Anthropic Claude Sonnet 4.6 serves as fallback when other providers fail.

• gpt-5.4 — OpenAIgeneration, chatAI models in use ↗
• gpt-5.4-mini — OpenAIauto-review, analysis, study-sheet generationAI models in use ↗
• gpt-5.4-nano — OpenAIclassification, fact-checkAI models in use ↗
• gemini-3-pro-image-preview — Google AIvisual diagram generationAI models in use ↗
• claude-sonnet-4-6 — Anthropicfallback when other providers fail (DEFAULT_MODEL)AI models in use ↗

Sub-processors

These are the vendors that process data on our behalf. All are bound by a Data Processing Addendum (DPA) compatible with GDPR.

• Cloudflare — Compute (Workers, Pages), object storage (R2), CDNGlobal edgeDPA ↗
• Neon — PostgreSQL managed databaseEU (Frankfurt)DPA ↗
• OpenAI — AI model inference (GPT-5.4 family)US (EU data residency available)DPA ↗
• Google AI — AI model inference (Gemini family, visual diagrams)GlobalDPA ↗
• Anthropic — AI model inference (Claude Sonnet 4.6, fallback model)USDPA ↗
• Resend — Transactional email deliveryEUDPA ↗
• Stripe — Payment processing (web checkout)Global (EU data residency)DPA ↗

Data Processing Addenda and certifications

Each sub-processor signs a GDPR Art. 28-compliant DPA. Links to public DPAs are available in the list above. Cloudflare holds SOC 2 Type II and ISO 27001 certifications. Neon stores data in EU regions (Frankfurt). AI models receive de-identified prompts without student identifiers (application-level anonymisation).

Audit trail and logging

For each AI model call we log: hex hash of the prompt (no content), model used, latency, tokens consumed, error code if applicable. The prompt content and response are NOT logged. Log retention: 90 days in Cloudflare Logpush + 30 days in PostHog (aggregated telemetry). Personally identifiable data (student name, teacher email) is NEVER sent to AI models — the system substitutes de-identified placeholders before invoking the LLM.

Your rights

Under the EU AI Act and GDPR, you have the right to be informed when content has been AI-generated, to request explanations about how the system works, to correct or reject any generated material, and to request the erasure of your data. To exercise these rights, contact us at contact@teachthemall.com.

Limitations and verification

AI models can produce incorrect, outdated or contextually inappropriate content. TeachThemAll applies automatic filters but does not guarantee the accuracy of all generated materials. It is the teacher's responsibility to review content before using it in the classroom.

Your role as an educator

As an education professional, you are responsible for verifying the pedagogical suitability and factual accuracy of materials before presenting them to your students. TeachThemAll is a support tool, not a replacement for professional teaching judgement. Your review is essential to ensure educational quality.

Contact and enquiries

For any questions about TeachThemAll's use of AI, to exercise your rights or to report a problem with generated content, write to us at contact@teachthemall.com. We respond within 72 working hours.