AI Services module
Contextual chatbot, AI-powered insights, mentor matching, and notification digests — backed by your choice of LLM provider.
AI Services adds a contextual chat assistant that can answer questions about your org's data, plus AI-powered analytics for org admins.
What this module enables
| Surface | What it does |
|---|---|
| Floating chatbot widget | Bottom-right of every page; ask questions in natural language |
| AI insights | Org admins get periodic summaries (e.g., "5 chapters at risk of falling behind on dues") |
| Document Q&A | Ask questions of your Documents library |
| AI mentorship pairing | (Optional) AI suggests mentor-mentee matches based on profiles |
How the chatbot works
- The user asks a question.
- GreekManage embeds the question and matches it against your org's data (members, documents, events, etc.).
- Relevant context plus the question is sent to a language model (Anthropic Claude, OpenAI, or Google Gemini — configurable).
- The model returns an answer, citing sources.
- The response streams back to the user.
This is retrieval-augmented generation (RAG). The model never sees data outside what's retrieved for the specific question.
Privacy boundaries
- Per-user scoping. A member can't get answers about another chapter's roster — the retrieval respects their access.
- No cross-tenant access. Org A's chatbot never sees Org B's data.
- Logging — conversations are logged for quality improvement; org admins can disable per-org.
- PII handling — the model receives only what's needed for the answer; sensitive fields (SSN, payment cards) are never sent.
Provider configuration
Org admins (or platform admins, depending on your setup) configure:
- Provider: Anthropic Claude, OpenAI, Google Gemini
- Model: provider-specific (e.g., Claude Sonnet, GPT-4, Gemini Pro)
- Knowledge scope: which data sources the chatbot can retrieve from (members, events, documents, invoices, etc.)
- Logging: on or off
- BYOK (bring your own key): paste your provider API key and pay the provider directly, or use the platform-managed key
Cost model
If you're using a platform-managed key, AI usage is billed via your subscription tier with a monthly query cap. If you use BYOK, you pay the provider directly.
Track usage in Org → Settings → AI → Usage.
Walkthroughs
| For | Page |
|---|---|
| Members using the chatbot | Using the AI chatbot |
Use cases
- "When does the next chapter meeting start?" — pulls from events
- "How much do I owe?" — pulls from invoices
- "What's the dress code for formal?" — pulls from event description and documents
- "Where can I find the bylaws?" — links to the document
- "Summarize last week's Engage discussions" — summarizes for org admins
Where AI can fail
- Hallucination — the model may confidently state something incorrect. Always verify financial or legal answers.
- Stale data — embeddings are refreshed on a schedule; very recent changes may not be retrievable for a few minutes.
- Privacy mistakes — if your data has been entered into the wrong access scope, the AI may surface it. Audit your access scopes.
Dependencies
- All other modules the AI is allowed to read from
- A configured provider key (BYOK or platform-managed)
- Storage — for embeddings (the platform manages this)
Pricing
AI Services is typically a paid add-on. Pricing depends on provider, model, and query volume. Confirm with your platform admin.
Related modules
- Documents — chatbot indexes documents for retrieval
- Operations — chatbot can answer dues, election, compliance questions
- Community — chatbot can summarize forum threads