Frequently Asked Questions

Gyanis AI Platform is an AI-as-a-service infrastructure built specifically for EdTech companies. It provides a single API to add intelligent tutoring, content generation, and conversational AI to any educational product — with built-in safety, student tracking, and analytics.

Gyanis AI adds an education-specific layer on top of foundational models. You get built-in content moderation, multi-entity management (students, teachers, institutions, classes, curricula), GDPR-ready data tools, webhook-driven event architecture, a tenant dashboard with analytics, and a training data pipeline — none of which come with raw model APIs.

Yes. Every AI response goes through automatic content moderation before reaching students. You can configure safety thresholds, receive moderation webhook alerts, and maintain full audit trails of student interactions.

All interactions are screened by an automatic moderation pipeline. Flagged content triggers webhook notifications. You have full visibility into moderation decisions through the analytics dashboard and can review flagged conversations.

Gyanis AI uses a prepaid credit model with two tiers. The Standard tier starts at $4.00 per million input tokens and $12.00 per million output tokens. The Premium tier is $6.00 per million input tokens and $22.00 per million output tokens. No monthly fees, no per-seat charges — pay only for what you use.

The platform includes GDPR-compliant data deletion tools, per-entity data export, full audit logging, and automatic PII stripping in training data. Data is encrypted at rest and in transit. The architecture supports data residency requirements for educational institutions.

Integration takes minutes. Create an API key in the dashboard, POST to https://api.gyanis.ai/platform/v1/chat with your entity_id, and you are live. The API is a standard REST endpoint — if you have used any chat API before, Gyanis AI works the same way.

All plans include access to documentation, API reference guides, and email support. For enterprise customers, dedicated onboarding, priority support, and custom SLAs are available.

Our models are trained on real educational interactions spanning JEE, NEET, UPSC, CBSE, ICSE, IGCSE, and IB curricula. This fine-tuning teaches the model pedagogical patterns — how to explain concepts step-by-step, when to ask clarifying questions, and how to adapt explanations to different student levels.

Retrieval-Augmented Generation retrieves relevant educational content from a curated knowledge base before generating each response. This grounds answers in verified material rather than relying solely on the model's training data, dramatically reducing hallucination and ensuring curriculum-aligned accuracy.

The multi-layer safety pipeline runs automatically on every interaction. You can configure safety thresholds per API key and receive webhook notifications when content is flagged. The pipeline includes input screening, context-aware filtering, output moderation, and continuous monitoring.

Yes. While our models are specifically fine-tuned for JEE, NEET, UPSC, CBSE, ICSE, IGCSE, and IB, the underlying educational AI capabilities work across any curriculum. Contact us to discuss fine-tuning for your specific educational context.

The input screening layer detects and blocks prompt injection attempts before they reach the model. Combined with context-aware filtering that enforces educational guardrails, the system prevents students from manipulating the AI into generating off-topic or inappropriate content.

Gyanis AI uses a standard REST API at api.gyanis.ai/platform/v1. The primary endpoints are /chat (synchronous) and /chat/stream (SSE streaming). Send a POST request with your API key and messages array. Additional parameters like entity_id enable entity tracking, and subject and grade provide educational context.

Most teams make their first successful API call within 15 minutes. Create an API key in the dashboard, POST to the /chat endpoint with your API key, and you are live. Use any HTTP client in any language — JavaScript fetch, Python requests, cURL, or anything else.

A simple POST request to https://api.gyanis.ai/platform/v1/chat with your API key in the Authorization header and a messages array in the body. Add entity_id for automatic entity tracking and conversation history. The response returns the AI message along with token usage stats.

Yes. Server-sent events (SSE) streaming is fully supported via POST /chat/stream. Each event is a chat.completion.chunk object. The final chunk includes token usage stats. A ping heartbeat is sent every 15 seconds to keep connections alive.

Any language that can make HTTP requests works with Gyanis AI. The API is a standard REST endpoint. Teams use JavaScript/TypeScript, Python, Java, Go, Ruby, PHP, C#, and more — anything with an HTTP client.

Two model tiers are available: mobius (standard) optimized for cost and speed, and oloid (premium) for complex topics requiring higher-capability models. The model tier can be set per-tenant or per-request. Model routing is admin-configurable with automatic failover via circuit breakers.

Every AI response passes through an automatic moderation pipeline before being delivered. Flagged content is held, a webhook notification is sent to your system, and the conversation is logged for review. You can view all moderation decisions in the analytics dashboard.

Yes. The platform is designed with K-12 safety as a first-class concern. Automatic content moderation screens every response. You can configure safety thresholds per API key, receive instant webhook alerts for flagged content, and maintain complete audit trails.

All entity data (students, teachers, institutions, classes, curricula) is encrypted at rest using AES-256 and in transit using TLS 1.3. Entity identities are isolated per tenant. Data access is controlled through scoped API keys with 11 granular permission scopes. Live and test modes keep development data separate from production.

Yes. A single DELETE API call to /entities/:id removes all data associated with an entity — including conversation history, metadata, and any training data. Associated request logs are automatically anonymized. This supports GDPR right-to-erasure requests and institutional data policies.

The platform is architected with FERPA requirements in mind. Entity data is encrypted, access-controlled, fully auditable, and deletable on demand. Per-entity data export supports data portability requirements. PII is automatically stripped from training data. Consult your legal team for your specific compliance obligations.

Average API response latency is under 200 milliseconds for non-streaming requests. Streaming responses begin delivering tokens within 100-300 milliseconds. Latency metrics are available in real-time on your analytics dashboard.

Gyanis AI provides a 99.9% uptime SLA. The platform runs on redundant multi-region infrastructure with automatic failover. Circuit breakers prevent cascading failures, and rate limiting protects your service during unexpected traffic spikes.

Gyanis AI supports 8 webhook event types: chat.completed, chat.error, credits.low, credits.depleted, usage.threshold, student.created, moderation.flagged, and taxonomy.updated. Every delivery is HMAC-SHA256 signed with headers X-Gyanis-Signature, X-Gyanis-Event, and X-Gyanis-Delivery-Id. Failed deliveries retry with exponential backoff: 30s, 2m, 15m, 1h, 4h (5 attempts max).

API keys support 11 granular scopes: * (all), chat, chat.stream, keys, usage, webhooks, entities (or students for backward compatibility), billing, compliance, taxonomy, and taxonomy.export. Keys can operate in live mode (production, billing enforced) or test mode (billing skipped). Scoped keys that lack the required scope receive a 403 endpoint_not_allowed error.

Yes. Per-key rate limiting with configurable RPM (requests/minute), TPM (tokens/minute), and daily request caps. Default is 60 RPM and 100K TPM. Rate limit headers (X-RateLimit-Limit-Requests, X-RateLimit-Remaining-Requests, X-RateLimit-Reset-Requests) are included in every response. Enterprise customers can request custom limits. Redis-backed distributed rate limiting is supported for multi-instance deployments.

All customers have access to comprehensive API documentation, integration guides, code examples, and email support at hello@gyanis.ai. Response times are typically within one business day.

Yes. Enterprise customers receive dedicated onboarding assistance, priority email support with faster response times, custom SLA agreements, and direct access to the engineering team for complex integration requirements.

Email hello@gyanis.ai with details about the issue or feature request. Include your tenant ID, API key prefix (never the full key), and steps to reproduce for bugs. Feature requests are reviewed and prioritized based on customer demand.