Credits
Credits pay for hosted agent runtime. Your model tokens are billed by your AI provider.
No subscription. Bring your own AI keys. Add credits and pay only for hosted agent runtime.
LLM token usage is billed directly by your model provider.
Model routing
Route the right model to each stage of one mission.
Give every stage a default model, override cards when a task needs different reasoning, and keep the work visible on one board.
Stage
Example model route
Reason
Research
GPT-class high-reasoning model
Explore ambiguity and produce implementation notes.
Execute
OpenRouter programming model such as Llama, Kimi, Qwen, or a custom model ID through codex_openrouter
Use searchable OpenRouter programming models where the catalog supports them.
Review
Gemini-class reviewer
Check behavior, coverage, and regression risk.
Security review
Opus-class reasoning model
Inspect sensitive changes before release.
PR description
Fast/small model
Summarize merged work without spending heavy reasoning.
Runtime control
Configure first. Add credits only for agent runtime.
Zero-credit workspaces can set up GitHub, BYOK providers, stage models, card models, MCP servers, and machine sizes. Runtime launches wait until credits are added.
Machines
Use low-cost shared Machines for routine stages. Switch heavy stages to performance CPU and more memory only when they need it.
BYOK
Connect OpenAI, OpenRouter, Claude, Gemini-class, or other supported provider keys without bundling model-token cost into Simple Kanban.
Product proof
Visuals for the launch page.
These launch captures show the actual Simple Kanban app with stages, cards, runtime credits, model routing metadata, run-history cost/spec details, and the session/chat workflow.
Free workspace
Create your control plane.
Claim a hosted workspace with zero runtime credits, configure the workflow, then add credits when the team is ready to run agents.