xpander.ai

xpander.ai is an enterprise platform for building, deploying, and governing AI agents that touch a company's full operational stack — CRMs, ITSMs, ticketing, finance, internal APIs, the lot. The product takes a deliberately unsexy position on the agent market: most demo agents fall over the moment they need to authenticate as a real user, paginate through a real API, survive a retry, or stay up for a week. xpander provides the surrounding scaffolding to fix that. Agents run on an always-on runtime; tool calls into 2,000+ pre-integrated systems carry per-user OAuth tokens; per-user context, memory, and state persist across sessions; and central governance layers — RBAC, audit logging, policy guards on tool calls, rate limits — apply uniformly whether the agent is invoked from xpander Studio, from custom code, or from an external MCP client like Claude Desktop or Cursor. Agents can be authored visually in xpander Studio or as code (Python and TypeScript SDKs), and routed to 80+ LLMs from OpenAI, Anthropic, Google, DeepSeek, or self-hosted endpoints. The MCP surface is bidirectional: xpander acts as an MCP server exposing its governed integrations to external clients, so a developer can drive xpander tools from Claude Desktop while the enterprise's auth, audit, and policies are still enforced centrally. Pricing as of June 2026 is published on xpander.ai/pricing: Cloud starts at $485/mo (5 users, 5 builders, 5 agents/workflows) with a 14-day free trial and no card; Self-Hosted on VPC, Kubernetes, or air-gapped infrastructure starts at $6,300/mo (50 users, 50 builders, 50 agents/workflows). Per-seat add-ons are User $19/mo, Builder $49/mo, and Agent/Workflow $29/mo. LLM token pricing is tiered (Premium models like Opus and GPT-5 Pro are $7.50 in / $40 out per 1M tokens; Standard $2.50 / $10; Economy $0.40 / $2), or teams can bring their own LLM tokens on Self-Hosted. Best fit: platform-engineering teams shipping production agents that have to survive contact with real users, real data, and real auditors.