Connect your tools

Living Context: custom connectors and field mapping

Living Context connects your AI agent to the systems where your customers actually act, such as your CRM, billing platform, or support desk. While standard connectors cover many common tools, you may need to go further. This article explains how to extend Living Context with custom connectors and MCP servers, and how field mapping gives you precise control over what the agent can see.

Connector types

Unless supports three ways to connect external data sources to Living Context. The right choice depends on the system you need and how it fits your stack.

Standard connectors

Pre-built connectors are available for common systems, including CRMs like Salesforce and HubSpot, helpdesks like Zendesk and Freshdesk, and knowledge bases like Confluence and Google Drive. These are configured with OAuth authentication and field selection through the Unless dashboard.

MCP servers

For broader integrations beyond a single endpoint, you can connect your own MCP (Model Context Protocol) server. This exposes internal tools, custom APIs, or proprietary data sources to the agent. The agent reads and acts through MCP the same way it does through any other connector, with the same audit trail. Available MCP integrations include Amplitude, Canva, Circleback, Digits, Render, Stitch, Superhuman, and Vercel, plus the option to build a custom MCP server.

Custom connectors

When your system is not on the standard list and MCP is not the right fit, you can build a custom connector using the Unless connector framework. This can be scaffolded by the Unless team or by your developers using the documented API. To get started, talk to your account manager.

How connectors work

When the agent needs data during a conversation, it queries the connector. The connector reads from your system using your credentials, returns the requested fields, and the agent uses them in the conversation. PII passes through the Privacy Vault before any model sees it.

Living Context is read live at conversation time, not pre-synced. There is no cached copy of your CRM inside Unless. If a connector is slow or unavailable, the agent degrades gracefully: it still answers, just without the Living Context that connector would have provided.

Field mapping

Each connector exposes data to the agent through field mappings. You decide which fields the agent sees, and you can limit what a connector exposes by editing the mapping. Only fields on the mapping list are available to the agent; everything else stays in your source system.

To see what a connector exposes, open the connector and click Inspect. The view shows sample data the agent will see during a conversation.

Tokenizing sensitive fields

In the field mapping, you can mark a field as sensitive. The Privacy Vault then tokenizes it. The agent sees a token, not the raw value, except when answering directly to the data subject.

When to use custom connectors or MCP

Consider a custom connector or MCP server when:

• Your system is not on the standard connector list
• You need the agent to see a field that is not in the default mapping
• An existing connector is exposing too much (for example, after a privacy review)
• You are setting up a new Moment that needs new data

Troubleshooting

If a connector returns no data, check three common causes:

1. The user used to authenticate lost permissions
2. The field mapping points at a field that no longer exists
3. The source system is unreachable

Conclusion

Custom connectors and MCP servers extend Living Context beyond the standard list, while field mapping keeps that extension under your control. Start with standard connectors where they work, and escalate to custom options only when your stack demands it. The same audit trail and privacy protections apply across all connector types, so you can extend without losing oversight.