A self-hosted AI gateway for small application teams
When an AI app needs several providers, one client-facing API and inspectable workflow runs, see where 1flowbase fits and how to evaluate it.
Read article1flowbase Blog
Start with a real problem.
Practical guides for AI application teams: evaluate a gateway, compose models, and investigate a run. Each article includes its limits and a concrete next step to test.
What are you working on?
Evaluate an AI gateway
Run the gateway evaluation checklistCompose models and tools
Explore the Fusion workflow guideWorkflow-backed virtual models: beyond LLM routing
A virtual model can be more than a provider alias: it can be a reusable, observable workflow composed from models and tools.
Read articleInvestigate an AI run
Run the review-app evaluation checklistFrom AI run logs to an internal review app
A practical 1flowbase pattern for teams that need to investigate AI runs, keep review decisions and build a small internal tool around execution evidence.
Read articleEngineering notes
Explore the documentationBuild a self-hosted feedback app with 1flowbase
Build a feedback app with 1flowbase data models and React. Keep observations, hypotheses, human decisions and status together, then verify persistence.
Read articleApprove agent changes without overwriting newer edits
A worked ticket example for keeping agent proposals separate from accepted data, rejecting stale approvals, and testing atomic writes, permissions and retries.
Read articleTurn agent conversations into a project tracker you can keep using
A practical pattern for turning confirmed decisions from agent conversations into project records, CRUD APIs and a React interface, with MCP for ongoing updates.
Read article