The AI Analytics dashboard (Dashboard → AI Analytics) gives administrators full visibility into how AI features are being used and what they cost.
What You Can Track
- Total AI spend — cumulative cost across all models and features
- Usage by model — which AI models are being used most (Gemini, GPT-4, Claude, etc.)
- Usage by feature — student chat vs. professor chat vs. document processing vs. voice grading
- Usage by user — identify high-usage accounts
- Volume over time — daily, weekly, and monthly trends
Understanding Cost Drivers
The largest cost drivers are typically: (1) student chat sessions with course agents, (2) voice session grading and analysis, and (3) document processing. Use the breakdown charts to identify which features drive the most spend.
Managing AI Costs
To reduce costs: select more efficient models from AI Models settings, configure agent answer reveal policies to reduce conversation length, and limit web search to courses that specifically need it.
Understanding Token Usage
Every AI interaction — whether a student sends a chat message, the system generates a quiz, or a resource is processed for embeddings — consumes tokens. A token is roughly 4 characters of text. Usage is billed by the underlying AI model provider (e.g., OpenAI, Anthropic) and passed through at cost. ScholarStack does not mark up AI costs.
The AI Analytics Dashboard — Metrics Explained
- Total Tokens Used — sum of input + output tokens across all AI operations in the selected period
- Estimated Cost — token count converted to USD at current model pricing. This is an estimate; actual invoiced amount may vary slightly.
- Tokens by Model — breakdown showing how much of usage came from each AI model (e.g., GPT-4o vs. GPT-4o-mini). Higher-tier models cost more per token.
- Tokens by Course — which courses are driving the most AI usage. High usage in a specific course may indicate engaged students or an over-reliance on AI for answers.
- Tokens by Operation Type — chat messages, document processing, quiz generation, summary generation, etc.