AI makes you faster. Is your plan worth executing?

AI can speed up execution. It cannot decide whether your business is pursuing the right outcome. Before you automate a task, define the outcome, challenge the assumption behind the plan, and decide what evidence would make you change course.

That is the starting point for Always Running. We have extraordinary tools available to us. But the discipline of running a business still matters: vision, judgement, a clear goal and a plan worth executing.

If the plan is flawed, faster execution can simply get you to the wrong place sooner.

Activity is not the same as progress

For a founder, “get this off my plate” is an understandable brief. There are enquiries to follow up, projects to check and decisions waiting for context. Handing work to AI can feel like the obvious next step.

The catch is that a task describes something to do. It does not necessarily describe the business outcome you need. Sending more messages, producing more reports or checking a project more often can create activity without resolving the underlying problem.

Before delegating the work, ask what would actually be different if it succeeded. Who would notice? What would they be able to do that they cannot do now?

An illustrative example: automating customer follow-up

This is a hypothetical teaching example, not a customer case study or a reported Walter result.

Imagine the brief is: “Get AI to follow up every lead, every day.” It sounds productive. There is an action, a frequency and a tool that might help carry it out.

But underneath that brief sits an assumption: more messages will mean more sales. What if the real problem is that nobody owns the next step? A prospect may have already replied, asked a question or agreed to a call. Another generic follow-up does not solve that coordination problem.

The first job is to improve the brief.

Replace the activity target with an outcome

A more useful goal is: “Every open enquiry has an owner and an agreed next action.”

That changes what you ask the system to do. Instead of maximising messages, you want it to help identify missing ownership, unresolved questions and the next useful action. You also need enough context to distinguish a genuine gap from a conversation that is already moving.

Keep the first test small. In this example, review ten open enquiries. Have a person check the suggested follow-up before anything goes out. For each enquiry, record its current position, owner and next action. Then review whether the action helped the conversation move forward—or merely added another message.

Ten enquiries will not prove that you have transformed sales. They give you a manageable way to inspect the plan, spot mistakes and decide what to test next.

Three checks before you ask AI to execute

1. What outcome do we want?

Describe a change in the business, not just an output. “Every enquiry has a clear next step” gives you something more useful to inspect than “send a follow-up”. Name who owns the outcome and what a satisfactory result looks like.

2. What are we assuming?

Write down what must be true for the plan to work. Perhaps the data is current, the customer has not already replied, or the team agrees who should act. A quick premortem helps: imagine the plan has failed, then list the most plausible reasons why.

3. What evidence would make us change course?

Decide this before the system starts. Duplicate messages, incorrect context or a missing owner might mean pausing and revising the workflow. Define what needs human review and when to stop, rather than treating every completed action as success.

Build a loop that improves the plan

The useful loop is simple: set a goal, run a small test, inspect the evidence and decide what changes. AI can help prepare and carry out the work. The founder still has to judge whether the direction makes sense.

Try this on one task in your business today. Write the outcome, the assumption and the evidence that would change your mind. Then give AI a plan worth executing.

Explore Always Running: visit AlwaysRunning.ai to learn more.