Skweel

Go beyond noticing that a query is slow. Work through the SQL, inspect the database context, and ask what a different approach could change.

An index recommendation is more useful when you understand why it might help. Bring questions about filtering, joins, and access patterns into the same place you write the query. Use the explanation to decide what to investigate, then validate the change against your database rather than treating an AI suggestion as the final answer.

Query plans and provider-reported metadata give the conversation something concrete to work with. Keep the path from a question to a candidate improvement short, without losing sight of the workload you are actually trying to improve.

Your database choice and your AI choice should not force you into a different workbench. Add the providers you need and keep the surrounding workflow familiar.

Work from the desktop application with its local server, connect that application to an organization-hosted server, or use the hosted workbench in a browser. Choose the arrangement that suits where your databases live and how your team works, without maintaining a separate client for each mode.

Database extensions expose the capabilities they support, from metadata to query execution and plans. Skweel keeps those differences visible rather than pretending every database works the same way. You can choose a provider for the job while keeping the rest of the workbench consistent.

Keep the conversation beside your work. Ask for an explanation, explore a tuning idea, or talk through the database before committing to the next query.

The companion belongs in the investigation, not in a separate chat window where you have to reconstruct the context from scratch. Work through a question, ask a follow-up, and examine the reasoning behind an answer while the relevant SQL and database remain close at hand.

For a wider view, open Schema Graph to explore the tables and relationships reported by your provider. Discuss the database from there, then open a table in Query when you are ready to work with it. Understanding the structure and writing the SQL become parts of the same task.

Use the AI service that fits your account, infrastructure, and preferences. Skweel supports provider profiles for cloud services, local Ollama models, and OpenAI-compatible endpoints.

Choose from Amazon Bedrock, Anthropic Claude, Azure OpenAI, GitHub Copilot, Google Gemini, Ollama, OpenAI, and compatible endpoints. Keep the provider settings in a saved profile instead of rebuilding the workbench around a single vendor.

Provider capabilities differ, including which tools a model can use. That choice remains part of your setup: select the provider and profile appropriate to the work you want to do, then keep the conversation within the familiar database workflow.

Install the database and AI extensions you need, manage their versions, and keep control over what is enabled.

The extension manager gives provider updates a place in the workflow. See which packages are installed, check their state, disable what you do not need, and use retained versions when a rollback is appropriate. Extending the workbench does not require starting over with another application.

You can also connect external AI tools through Skweel's MCP access workspace. Issue a revocable credential for one saved database profile and choose the available read-only capabilities it should expose. Give a tool the context it needs without handing it an unrestricted connection configuration.