Plan your setup
Choose the first workflow, where it runs, the model services and an ongoing budget.
Dude is our example, not a template for your life. Build an assistant around your tasks, with accounts and controls that stay yours.
You will need a computer or server, a model account or suitable local model, and time to configure, test and maintain the setup. Some command-line confidence helps.
Write down the input, desired result and what a successful run looks like. Start with a research briefing or another task you can easily check.
Compare a local machine with a hosted server. Decide which data may go to a cloud model, and estimate model usage, hosting and service subscriptions.
Read the architecture guide →Use accounts you own and grant only the permissions required for that first task. Keep credentials out of prompts and public logs.
Explore permissions and approval →Check real outputs, failure cases and costs. Test how to pause the task and remove access before adding a schedule.
Explore operating patterns →Add another task only after you understand the first. Keep a record of what works, what needs review and what should remain yours to do.
For individuals and small teams who want help getting started. We discuss your first use case, then agree the scope and price before any paid work begins.
Choose the first workflow, where it runs, the model services and an ongoing budget.
Configure the agreed tools and approval rules, then test the workflow with examples you can review.
Get a walkthrough and setup notes so you can operate, pause and adapt your assistant. Any ongoing support is agreed separately.
Model subscriptions, hosting and third-party fees are separate from the setup work. Important outputs and actions still need your review.
Discuss your setupOur apps, experiments and Dude’s own blog show one possible direction.