Design a CLI that lets an agent complete a real SaaS setup, discover the next action from useful command output, and cross an email verification step when the user has granted the necessary tool access.
Source episode · 56:18
How Intercom Doubled Engineering Output: Brian Scanlan's 4 AI Workflows for Claude Codewith Claire Vo
Brian asks an agent to install Intercom through a CLI, uses helpful command output to suggest the next action, and lets the agent retrieve an email verification code through an available tool.
An agent can create an account, complete verification, configure the product, and report the resulting installation state through the CLI.
Anthropic AI assistant
Step by step
Follow the sequence once, then adapt the prompts, checks, and handoffs to your own setup.
4 steps
Choose one valuable end to end task, such as installing the product on a site. Write the outcome from the user’s perspective before designing commands.
Install Intercom on my website. Create the account, complete the required setup, add the installation snippet, and report the workspace and installation status.
Expose each required operation through clear CLI verbs and stable flags. Support structured output, idempotent retries, status commands, and errors that identify the exact unresolved state.
When a command cannot continue, make its help text useful to an agent. State what happened, what input is missing, and the exact next command. For email verification, suggest checking the authorized inbox and provide a command that accepts the code.
Verification required. Retrieve the code sent to the account email, then run `intercom auth verify --code <code>`. If you have authorized email access, search for the latest Intercom verification message.
This hint bridges the gap between automated steps and those that typically require a human.
Give the agent only the adjacent tools needed for the task, such as an email client or browser, with explicit user authorization. Test the entire flow from a clean account and confirm the final product state rather than trusting a successful command exit.
Turn an idea into a PRD, user stories, and a plan.
After the steps
How to recover when the loop fails and where human judgment helps.
Recover
Keep building
Log your workouts effortlessly without ever opening a spreadsheet. This workflow uses the Claude mobile app to parse a simple voice note and automatically update your workout tracking spreadsheet in the background.
Create a custom email management interface within Claude to read, triage, and draft replies without opening the Gmail tab. This workflow helps you reclaim focus and compound your learnings by training your AI on every interaction.
Build a fully automated system in Claude to generate personalized, interactive HTML proposals and client onboarding materials, saving hours on administrative work and creating a high-touch experience.
Join 100,000+ product managers who use ChatPRD to write better docs, align teams faster, and build products users love.