Frequently Asked Questions About xpander.ai
What is Smart Engage and how does it save costs?
Smart Engage is xpander.ai’s adaptive inference scheduler—it analyzes message sentiment, thread depth, user status (e.g., “away”), and task urgency to determine *if*, *when*, and *how much* LLM compute an interaction truly requires. By suppressing low-value calls, consolidating follow-ups, and caching deterministic responses, it slashes token consumption while maintaining conversational fidelity—turning unpredictable LLM bills into fixed, scalable line items.
How does user authentication work in xpander.ai?
Authentication is user-centric and consent-driven. When a Slack user first interacts with your agent, they’re guided through a single, branded OAuth flow. xpander.ai securely stores scoped tokens (e.g., repo:read, jira:issue:write) tied to that user—never the workspace. Subsequent requests execute *in their context*, ensuring data isolation, audit trails, and compliance with least-privilege principles across all integrated platforms.
Can I use my existing AI agents with xpander.ai?
Absolutely. xpander.ai’s runtime accepts standard HTTP handlers (Flask/FastAPI), LangChain chains, CrewAI agents, or raw Python scripts. With minimal configuration—mainly specifying input/output schemas and auth scopes—you retain full ownership of your logic while inheriting xpander.ai’s Slack integration, memory layer, cost controls, and observability. Refactoring is optional—not required.
How long does deployment take with xpander.ai?
From idea to live agent: under 7 minutes. Create a blueprint or upload code → connect Slack + select channels → enable OAuth (if needed) → hit “Deploy.” Your agent appears in Slack immediately, responds to /xpander help, and begins processing messages—no waiting for builds, no manual webhook setup, no infrastructure provisioning. Iteration is just as fast: edit, save, and redeploy in seconds.