Eternal Loadstar Journal

Ideas to putto work.

Practical notes on better systems, useful technology and the people who use them. Start with a question from your own working day.

Eternal notes

01 / Systems

A useful starting point:
one task, from beginning to end.

Before you build,
map the work.

A request arrives by email. Someone copies it into a spreadsheet, checks a second tool and asks a colleague what happens next. The obvious answer might seem to be a new app. First, write down what the work actually involves. A simple map can reveal whether you need software, a clearer decision or a step removed.

Follow one real task.

Choose a task your team repeats, such as handling an inquiry or approving a purchase. Start with its trigger and finish with a specific result. Ask someone who does the work to walk through a recent example. Record the steps in order, including the waiting, copying and checking that rarely appears in a formal process.

For each step, note the person responsible, the information they need, the tool they use and what they pass on. Mark places where information is missing, access is difficult or the same details are entered again. Keep a separate note of what happens when the usual path breaks.

Choose the change before the tool.

Ask which obstacle matters most. Does a request need a named owner? Could one shared record replace several copies? Would a clear approval rule remove a recurring delay? These questions give a future system a job to do, rather than a list of features to accumulate.

Describe the proposed change in one sentence, then decide how you will check it. For example: a request should reach the right person with the required details, including when something is missing. Test that path with the people who will use it before adding more scope.

You now have the beginning of a useful brief: the task, its users, the problem, the information involved and the result to demonstrate. That is a stronger starting point for building than choosing a platform first.

How we build business systems

Eternal notes

02 / Learning

A useful habit:
try, check, improve, repeat.

Learn AI with
a task in mind.

Learning an AI tool is easier when the exercise has a clear purpose. Pick a small task you understand well: turning fictional meeting notes into action items, simplifying a sample instruction or preparing questions for a project brief. Familiar work gives you a way to judge the answer. An impressive response is not automatically a useful one.

Make the exercise repeatable.

Write down the input, the intended audience and the output you need. Tell the tool what to preserve and what to leave uncertain. For an action list, that could mean keeping the stated owners and dates, and marking missing details instead of filling them in. Use fictional or approved sample material while you learn, especially when workplace information is involved.

Keep the original input beside the result. Check names, numbers, omissions and any statement that was not supplied. Ask whether another person could use the output without guessing. If an answer sounds certain where the source is unclear, revise the instruction and compare the next attempt.

Practice the check, not just the prompt.

Change one part of the instruction at a time. Save a useful example with a short note explaining what worked and what still needed correction. Then try a second example with incomplete information or an unusual request. This shows whether the method survives beyond the easiest case.

Agree where human review belongs before using the method in daily work. A draft can help someone think or prepare; sending it, changing a record or making a decision is a separate step. The person responsible should know what they are approving and what evidence they need.

A good learning session ends with a repeatable task, a checking routine and a clear limit. Build confidence through those small, understood steps before expanding what the tool is asked to do.

Explore learning and skills