AI Application Platforms
AI application platforms combine model access with user interfaces, visual workflows, agents, tools, retrieval, and application state. They let teams build and operate AI experiences without assembling every component directly in application code.
When a platform accepts an OpenAI-compatible endpoint, its language-model requests can pass through AISIX for caller authentication, model aliases, routing, policy, and telemetry. The platform continues to run the application workflow and user experience, while AISIX governs the request path from the platform to the configured model provider.
Choose a Platform Guide
Each platform exposes the AISIX connection through a different configuration surface:
| Platform | Integration surface | Guide |
|---|---|---|
| Open WebUI | OpenAI-compatible connection | Open WebUI |
| n8n | OpenAI Chat Model sub-node | n8n |
| Dify | OpenAI-API-compatible model provider plugin | Dify |
These platforms are AISIX clients, not upstream model providers. Configure provider credentials and upstream model IDs in AISIX. Give the platform an AISIX caller API key and model alias instead.
Prepare AISIX
Each platform needs these gateway-facing values:
- An AISIX proxy URL the platform deployment can reach.
- An AISIX caller API key dedicated to the platform or application.
- A chat-capable model alias that the caller key can access through
POST /v1/chat/completions.
If AISIX is already deployed for your organization, obtain these values from the team that manages it. Otherwise, follow the Open-Source AISIX Gateway Quickstart or AISIX Cloud Quickstart, or contact API7 for Hybrid Cloud access.
Use a caller key scoped to only the aliases the platform needs. A shared platform connection normally identifies the platform or application to AISIX, not each person using its interface. Keep end-user authorization and audit context in the application platform unless you have separately designed and verified per-user identity forwarding.
Understand the Boundary
AISIX receives model requests after the application platform has assembled prompts, retrieved context, or selected tools. AISIX does not replace the platform's workflow engine, user accounts, conversation state, knowledge base, vector store, or tool execution.
Gateway policies apply to the supported content that reaches AISIX. A platform may perform additional model calls for titles, summaries, memory, retrieval, or agent planning. Confirm which calls use the configured AISIX connection before relying on gateway telemetry or limits as a complete record of the platform's activity.
AISIX model discovery returns every alias the caller key can access, including aliases intended for non-chat endpoints. Select an alias configured for Chat Completions in each application platform.
Tool calling also crosses both product boundaries. AISIX must preserve the model's OpenAI-compatible tool-call response, and the application platform must execute the tool and send the result in a follow-up request. Test a complete tool round trip instead of treating a successful text response as proof of tool compatibility.
Verify an Integration
Use one short platform workflow to confirm the complete path:
- The platform sends a prompt through the configured AISIX connection.
- AISIX records
POST /v1/chat/completionsfor the expected caller key and model alias. - The platform renders or processes the model response.
After the first success, verify streaming, tools, retrieval, and background model calls that the application depends on.
Related Guides
Choose a platform guide above for setup. These guides explain the gateway behavior shared by the integrations:
- OpenAI-Compatible API: review the gateway-facing request format.
- Streaming: understand streamed response behavior.
- Tool Calling: verify complete tool-call loops.
- API Key and Model Rate Limits: limit platform traffic.