OpenAI provider setup

Nexaflow agents need an AI provider to generate responses and, depending on configuration, process knowledge search.

Use this guide when your deployment expects an OpenAI API key.

#What to configure

Set the OpenAI API key in the server environment where Nexaflow runs.

Common environment variable:

OPENAI_API_KEY=sk-...
bash

Do not expose this key in frontend code.

#After configuring the key

1. Restart the backend process.
2. Open the app.
3. Create or open an agent.
4. Select an available model in Agent Settings.
5. Test in Playground.

#Model selection

Agent settings may show several model options. Choose based on the workflow:

• Smaller models for basic FAQs and routing.
• Balanced models for support and lead qualification.
• Stronger reasoning models for complex policy, operations, or multi-step workflows.

Use Playground and Analytics to compare quality and cost.

#Troubleshooting

The agent does not respond

Check:

• OPENAI_API_KEY is present in the backend environment.
• Backend was restarted after adding the key.
• The selected model is supported by your deployment.
• Workspace has available credits.

Knowledge upload fails

If embeddings use OpenAI in your deployment, missing provider credentials can prevent knowledge indexing.

#Related guides

• Configure agent settings
• Grounding basics