Overview
OpenWebUI is a web-based interface for interacting with large language models through a conversational workspace. In a Shakudo environment, OpenWebUI gives customers a user-friendly chat experience while Shakudo manages the underlying infrastructure and approved model access path.
How OpenWebUI fits into Shakudo
- Customers launch OpenWebUI from Stack Components in the Shakudo platform.
- OpenWebUI is commonly connected to LiteLLM so teams can use one interface for approved model endpoints and routing policies.
- Model providers such as Azure OpenAI, OpenAI, Anthropic, or other supported backends are configured through the approved Shakudo setup path.
- Users focus on chatting, testing prompts, and collaborating while Shakudo handles deployment, networking, and access boundaries.
What customers use OpenWebUI for
- Chat with approved LLMs through a clean workspace interface.
- Compare model behavior for prompt testing and exploration.
- Support internal copilots and lightweight team knowledge workflows.
- Collaborate on prompts, conversations, and evaluation patterns inside a shared platform environment.
OpenWebUI, LiteLLM, and model providers
- OpenWebUI is the chat interface.
- LiteLLM often acts as the gateway or proxy layer for model routing, credentials, and standardized access.
- The actual model providers supply the underlying inference endpoints.
- This separation helps customers manage model choice and provider access without changing the end-user chat workflow.
What Shakudo manages
- Component deployment and service availability.
- Platform access, routing, and workspace context.
- Approved integrations and environment-level configuration.
- Operational support for the managed Stack Component path.
Getting Started
Start from Shakudo. Customers should use the managed OpenWebUI component inside the platform rather than trying to access a separate self-hosted OpenWebUI instance directly.
1. Access OpenWebUI from Shakudo
- Log in to your Shakudo workspace with your approved account.
- Open Stack Components from the Shakudo navigation.
- Select OpenWebUI from the available component list.
- Launch the component and wait for the chat workspace to open.
2. Confirm available models
- Check which models are exposed in the model selector.
- If your workspace uses LiteLLM, the list may reflect centrally approved models routed through the LiteLLM gateway.
- If expected models are missing, confirm whether they were enabled for your workspace or environment.
3. Configure model access if required
- Some environments provide preconfigured models, while others require an administrator to connect approved model-provider credentials first.
- Use only the supported provider and credential path defined by your Shakudo setup.
- Avoid entering personal or unapproved production credentials into shared team environments.
4. Start chatting
- Pick an approved model from the selector.
- Start with a simple prompt to confirm the chat path is working.
- Use short iterative prompts first before moving to longer team workflows.
- Compare responses across models when you need to evaluate quality, latency, or style.
5. First validation checklist
- OpenWebUI launches from Shakudo successfully.
- At least one approved model is visible.
- A test prompt returns a response.
- The user understands which models and workspaces are approved for their team.
Administration and Best Practices
OpenWebUI is easiest to manage well when access, model availability, and usage expectations are clearly defined at the workspace level.
User management
- Limit access to approved users or groups.
- Separate test users, power users, and production-facing user groups when needed.
- Review access regularly as teams and responsibilities change.
Model management
- Expose only the models that are approved for the workspace or use case.
- Prefer a centralized routing layer such as LiteLLM when you want consistent model access and easier credential management.
- Document what each available model is intended for so users can choose appropriately.
Conversation practices
- Do not paste regulated, secret, or unapproved customer data into prompts unless the environment and policy explicitly allow it.
- Use repeatable prompt templates for common workflows instead of relying only on ad hoc chats.
- Periodically review how teams are using the component so useful patterns can be documented and repeated.
Operational best practices
- Track recurring login, model visibility, or latency issues and turn them into runbook updates.
- Keep provider credentials and gateway configuration out of user-facing chat instructions.
- Use production change control for model-provider, gateway, or access-policy changes.
Troubleshooting and FAQ
Problem: I cannot log in or OpenWebUI does not open
- What to check: Confirm you are logged into the correct Shakudo workspace and that OpenWebUI is enabled in Stack Components.
- What to check: Verify your Shakudo session has not expired.
- Resolution: Relaunch the component from Shakudo. If the issue persists, ask your administrator to confirm access and component availability.
Problem: Expected models are missing
- What to check: Confirm which models are approved for your workspace.
- What to check: If the environment uses LiteLLM, verify the gateway configuration includes the expected models.
- Resolution: Ask your Shakudo administrator to enable or expose the required models through the approved model access path.
Problem: Responses are slow
- What to check: Determine whether the issue affects one model or all models.
- What to check: Review whether provider latency, gateway routing, or workspace load may be contributing.
- Resolution: Retry with a smaller prompt, test another approved model, and escalate with timing details if latency persists.
Problem: Chats work in one workspace but not another
- What to check: Compare model availability, permissions, and component enablement across workspaces.
- Resolution: Align the workspace configuration or use the environment intended for that workflow.
Problem: Who should I contact for support?
- Collect the Shakudo workspace, approximate time of issue, selected model, and any visible error message.
- Provide the component name and the steps taken before the problem appeared.
- Share this information with your Shakudo administrator or support contact for faster triage.
