API Overview

Jardine’s API surface is the foundation beneath the dashboard and the extension point for deeper operational automation. Even if you are primarily using the UI, understanding the API shape helps your team reason about behavior, investigate issues, and design integrations that stay maintainable.

A useful mental model is that the API follows the same support system layers as the product: identity and scope, channel setup, knowledge, routing policy, runtime context, and operational outcomes.

Request Model and Scope

Most endpoints are under /api/* and are scoped to organization context. In the frontend client, requests are sent through a shared axios instance that can attach context headers like organization and workspace identifiers when available.

That scope behavior matters. It ensures that actions like connector updates, routing changes, and knowledge ingestion are applied in the correct tenant context.

For teams building automation around the API, always treat scope as explicit. Never assume defaults in production workflows.

Core API Domains

The API is easiest to understand in groups.

Authentication endpoints handle user identity and session lifecycle: register, login, profile fetch, profile updates, password reset flows, and OAuth exchange.

Organization endpoints handle org identity and settings, including support behavior settings that drive onboarding completion state.

Workspace endpoints cover workspace listing and people-management style operations such as listing members, inviting members, role updates, and removals. In current product behavior, organizations run in single-workspace mode for operations, but workspace APIs still exist and are used by team management flows.

Knowledge endpoints cover collections, documents, ingestion flows, ask behavior, and analysis data.

Validation endpoints cover playground conversations, message creation, query ratings, suite management, suite runs, and outcome simulation.

Conversation endpoints expose conversation listing, detail retrieval, and handover actions between AI and human ownership.

Routing endpoints cover tags, destinations, and rules.

Connector endpoints cover connector lifecycle, validation, schema snapshot, credential rotation, templates, template testing, and Stripe-specific connect flows.

Channel and integration endpoints cover email channel setup and provider-specific integrations like Intercom and Zendesk.

Why This Grouping Helps in Practice

When incidents happen, debugging is faster when you can map symptoms to API domain.

If inbound traffic is missing, investigate channel and webhook setup endpoints first. If answers are weak, investigate knowledge and validation endpoints. If escalations feel noisy, inspect routing endpoints. If account-aware behavior is broken, inspect connector and template endpoints.

This layer-first approach prevents random changes and shortens time to root cause.

Designing Integrations Against the API

Teams extending Jardine through automation usually get best results when they adopt three principles.

First, keep workflows idempotent where possible, especially around ingestion and routing updates. Retried jobs should not create accidental duplicates.

Second, design for observability. Log request intent, endpoint group, and scoped identifiers in your integration layer so troubleshooting remains clear.

Third, separate setup-time automation from run-time automation. For example, connector and routing changes should be treated as controlled config updates, not dynamic runtime improvisation.

These patterns reduce operational risk as your integration footprint grows.

Common Integration Mistakes

One common mistake is coupling too many product areas into one automation job. A single script that edits knowledge, routing, and connectors in one run can make rollback and debugging difficult.

Another mistake is assuming “successful HTTP call” means “complete.” Many workflows, especially around ingestion and validation, require follow-up checks.

A third mistake is ignoring human process boundaries. API automation should support support teams, not bypass their policy controls.

Final Guidance

You do not need to call every endpoint to get value. Most teams benefit from using the dashboard for baseline operations and using API automation for repeatable, high-impact tasks where consistency matters.

If you keep your integration design aligned with support layers, scoped correctly, and observable by default, the API becomes a multiplier rather than a maintenance burden.

That is the goal: reliable extension without operational drift.