Connect

85+ connectors, multi-model routing, open interop

85+ connectors bring your data in fast and governed, backed by an engine that sustains 2,875 writes a second even on local disk, so your estate streams in without batch windows.

Connectors

Wire your entire estate

CRMs, collaboration, mail, ticketing, engineering, databases, messaging, voice, web, and search. Each exposes governed operations with risk, scope, and classification metadata, streaming into the engine at up to 5,588 writes a second in-memory.

SlackSlackCommunication
GmailGmailCommunication
TeamsTeamsCommunication
SalesforceSalesforceCRM
HubSpotHubSpotCRM
JiraJiraProductivity
NotionNotionProductivity
LinearLinearProductivity
SAPSAPEnterprise
ServiceNowServiceNowEnterprise
MMongoDBData
PPostgreSQLData
ClaudeClaudeAI Models
GGPTAI Models
GeminiGeminiAI Models
MMistralAI Models
+65 more connectors

How connectors work

Every connector exposes governed read/write operations with classification metadata

Connection lifecycle

1.

Configure

Provide credentials (encrypted, stored in your KMS)

2.

Discover

Connector auto-discovers available operations and data shapes

3.

Classify

Data classification applied at ingestion (public, internal, confidential, restricted)

4.

Govern

Every operation wrapped in policy-checking and audit

5.

Expose

Operations available as governed tools for agents

Data management

You decide what each connector feeds the graph

Ingestion is not a black box. Every source is controlled, every mapping is reviewable, and schema changes surface before they break anything.

Ingestion control

Per-connector control over what flows in. Pause, resume, and scope each source independently.

Sensitive sources onboard at your pace, with scope you set.

Field-level mapping review

Review how source fields map into the graph, field by field. Approve, correct, or exclude before data reaches agent context.

Bad fields are excluded before they ever reach an agent prompt.

Schema-drift detection

When a source schema changes, the drift is detected and flagged. Resolve it in one place, without breaking downstream context.

Upstream changes become review tasks, not silent context corruption.

Multi-model registry

No hard-coded AI provider. Route by data classification, cost, latency, and sovereignty.

The registry supports hosted and bring-your-own models, each fed 2,668-token context payloads. Routing rules select models by data class, context size, risk, or department.

Sovereignty routing

Classified data never leaves your boundary. Restricted classes pinned to local models automatically.

Cost optimization

Route by economics. Per-organization credits, quotas, and projected-spend visibility.

Model freedom

Claude, GPT, Llama, Gemini, Bedrock, and local runtimes. Swap without code changes.

Fallback chains

Cost ceilings and fallbacks balance sovereignty against latency and cost.

Supported providers

  • Anthropic (Claude family)
  • OpenAI (GPT family)
  • Google (Gemini family)
  • Meta (Llama family)
  • Mistral
  • AWS Bedrock
  • Azure OpenAI
  • Local runtimes (Ollama, vLLM, TGI)

Routing rules

Restricted data

Local models only (sovereignty enforcement)

Cost optimization

Route by token price per classification level

Latency requirements

Route by response time SLA

Capability matching

Route by model strengths (reasoning, coding, summarization)

Fallback chains

  • Primary model unavailable, automatic failover to next in chain
  • All cloud models down, fall back to local runtime
  • Sovereignty constraints always respected during failover

Open interop

MCP server, MCP client, A2A federation

Simultaneously an MCP endpoint for inbound agents and an MCP/A2A client to outbound platforms. Every direction, inbound or federated, reads the same engine.

MCP Server

Wexa exposes all governed operations as MCP tools

  • Any MCP-compatible client can discover, invoke, and observe through Wexa
  • Governance wraps around every inbound operation

MCP Client

Wexa connects to external MCP servers

  • Federate to third-party services with full governance on outbound operations
  • Tool discovery, invocation, and audit, all governed

A2A Federation

Agent-to-Agent protocol support

  • Capability discovery handshake between platforms
  • Cross-platform agent orchestration with governance boundary enforcement
  • Every federated interaction audited and policy-checked

Architecture

Governed interoperability at a glance

MCP INBOUNDAgent AAgent BAgent CWexaGOVERNED CONTROL PLANEServiceNowSalesforceCustomMCP / A2A OUTBOUND
Get started

85+ connectors. Connect your stack today.

Tell us which systems you need to connect. We'll show you exactly how Wexa integrates with your environment.

Book a Demo Contact us