A free tool from The Alinement Brief

The AI Workflow Card

Three rows for the tasks AI is allowed to touch in your business: what it does there, who answers for the output, and the line it never crosses. Fill it in with your leadership team in one sitting, then keep it where your scoreboard lives.

Before you start

Start with tasks that drain time, not headlines.

Don't start with "what can AI do." That question produces endless prompts in ChatGPT and Claude. Start with "what drains hours every week that is mostly information handling." That question provides a good starting point and focus: sales quote follow-ups, meeting recordings turned into notes and action items, first drafts of routine customer replies, job postings, sorting a week of inbound email. None of it makes headlines on LinkedIn or in YouTube videos. That's the point.

Pick your three candidates before you touch the table. If a task doesn't pass the three checks printed on the card, it doesn't earn a row, no matter how impressive the demo was.

This quarter's card

One workflow per row.

A task earns a row when it passes three checks: it happens every week. It is information handling. And faster would matter to the customer, the team, or your own work.
The task What AI does (one sentence) Owner (one name) The guardrail ("never...")
Three rows is the whole design. When a row lasts for a quarter, incorporate it into how that job is done going forward, owner and guardrail included, and free the slot. When a row keeps producing output its owner rewrites every time, stop it. The card is a guide to responsibly leveraging AI in your operations.
Two house rules sit above every row. The data rule: if you wouldn't email it to a stranger, don't paste it into a tool nobody vetted. And even in a vetted paid tool, a row touches only the specific data it names, uses only what the task needs, with the owner's sign-off. The review rule: nothing AI wrote leaves the building until its owner has read every word and checked any fact, figure, name, or link it claims.

Column two sets the boundary

AI drafts. A person decides.

Write one sentence for what AI does in this workflow, and watch the verbs. AI drafts, summarizes, sorts, compares, and flags information. A person decides, approves, sends, or signs off on next steps. If your sentence ends with the work bypassing human review, rewrite it until a person is in the loop.

"Drafts the follow-up email from the quote and the site notes" is a workflow. "Handles our follow-ups" is an unsupervised employee nobody hired.

Column three is the same rule as always

One name answers for the output.

Not a department, and not "whoever ran the tool that day." The owner is not necessarily the person who types the prompts. The owner is the person who is responsible for the output, reads it before it moves forward, and tells you when the workflow should be promoted, fixed, or stopped. A workflow without an owner isn't automation. It's unsupervised work.

Column four draws the line

Every guardrail starts with "never."

One sentence per row, and it usually takes one of two forms. A data condition: what never gets pasted in. Customer lists, employee records, payroll, banking details, anything in an NDA. Or an output line: what never goes out without the owner reading every word. Write the one guardrail that fits the task. If nothing comes to mind, you haven't thought hard enough about the task yet.

This card is the one-page version of an argument The Brief has been making since Issue 3. The full walkthrough is in Issue 12 of The Alinement Brief; the argument behind it is Issue 3. Once your three rows are filled in, keep them honest inside the weekly rhythm: the Priority Scoreboard for status, the 60-Minute Agenda for decisions. Or browse every free tool.

Get the next tool in your inbox