Application Deployment and Lifecycle
How to manage the lifecycle of an AI application
13 articles
- Pause and restart a thunkStop all AI work on a thunk and start it again from the Monitor pane
- Move a thunk to a new AI modelTry a different AI model on your test work items, compare results, cost, and time, then switch the thunk over safely.
- From prototype to production: the thunk lifecycleHow a thunk goes from design to test, review, promotion, monitoring, and upgrade — and which article covers each stage.
- How to save a report from your work itemsSave a filtered, column-trimmed view of your work items as a named report you can reopen and share.
- Lifecycle of a ThunkMove your thunk through Prototype, Testing, and Production
- Observability and MonitoringSend a webhook to your systems when a work item finishes
- Supported AI models and what happens when they changeWhich AI model to choose for your thunks, which models are available today, and what happens when a model is phased out
- Time Analysis: how long your thunk takesSee how long a work item takes to get through your workflow — or through a single step — and which specific items were the slow ones.
- Token Costs: what your thunk spends on AISee what a thunk spends on AI — day by day, for the whole workflow or for a single step — and what it typically costs to put one work…
- Troubleshooting ErrorsA complete reference to every error Thunk can raise: what each one means, why it happens, and what to do next — organized by error category.
- Understanding LLM costs and latencyLearn how the LLM models used by AI agents incur cost and delay
- Versioning Support for your ThunkLearn how Thunk automatically tracks every change to your workflow definitions, so you can see what changed, when, and restore previous values if needed.
- Workflow ReviewCatch thunk design issues before they cause production failures with a static analysis review of your workflow configuration and step instructions.