You will bring together everything you built into one kit you will use every week and can share with your team.
Open a new document or folder and name it your AI work kit. Under the name, write down where each piece of your course work is stored at the moment.
Most learners reach this point with their work in several places. Your voice guide might be in the assistant's custom instructions, your meeting summary prompt in a project, your research checklist in a notebook, the data rules in an email from your manager, and your time-saved figures in your log. Each of these works fine by itself. Kept apart, though, they are hard to find and impossible to share, and it is easy to let them go stale without noticing. The capstone project collects them in one place.
A kit is one document or folder holding everything you need to repeat what worked in the course. It holds four kinds of material: prompts and guides for each kind of task, data rules, results, and a review date. You are writing it for two readers. The first is yourself in three months, once you have forgotten the details. The second is a colleague who has not taken the course. You already have most of the material, so the job is to collect it, tidy it and fill the gaps. Allow about an hour.
Find the working version of each prompt and guide and give each its own section. Use the latest form of your voice guide from lesson 3.1, with any lines you added after the exercise in lesson 3.4. From module 4, bring your research checklist: the steps for tracing a claim to its source, the four labels for marking claims, and the one-page brief structure. From module 5, bring your formula prompts: the layout description you give the assistant, the request for an explanation of each part, and a note on adding a check cell. Add the slide storyline prompt from lesson 6.1, including the instruction for sentence headlines, and the meeting summary prompt from lesson 7.2 with its four sections and its rules on gaps and timestamps.
Under each one, write a line saying when to use it and one thing to watch for. Someone else can only use your kit if these lines are there. A model line: "Meeting summary prompt, use after any meeting where I own the follow-up, watch for suggestions recorded as decisions."
The data rules section tells the reader which data may go into which tool. List the company-approved tools from lesson 2.3, and next to each one write the kinds of data allowed. Describe the redaction method from lesson 2.4 in a few lines: placeholders, rounding, removing what the task does not need, and keeping the key outside the chat. If a rule came from a conversation and not from a written policy, record who you asked and when you asked them. Then add the one-sentence recording notice you wrote in lesson 7.1.
Keep the section to one page at most. A colleague should be able to read it in two minutes and know what they may paste where.
Add your time-saved figures from lesson 9.1 task by task. For each task, give the baseline, the new average, the net saving per week and the label: saving, no change or loss. Include any quality notes that mattered, for example errors that a check caught.
After the results, add a short list of exceptions: the tasks you chose to keep doing yourself, with one sentence of reasoning for each. People often leave this list out. It is one of the most useful parts of the kit, because it saves others the time of finding the same limits for themselves. Priya, an invented example learner, has two exceptions in her kit. She does the monthly regional update by hand because the assistant consistently misread the figures. She does supplier invoice reconciliation by hand because checking the assistant's work would take as long as doing it herself. Anyone who picks up her kit avoids spending a month rediscovering those limits.
AI tools, company policies and your own work all change over time, so a kit that was accurate in October can be misleading by the following year. Put a review date three months from now at the top of the kit, and put the same date in your calendar. When the date comes, check that each prompt still works, that the data rules still match policy, and that each task is still worth doing with AI.
Then share the kit with one colleague who does similar work. Ask them to try one prompt on a real task and tell you what was unclear. Their questions point to the places where the kit relies on knowledge only you have. Priya's finished kit was four pages long. When she shared it, her colleague mainly asked where the email prompt was kept, so she added a line naming the exact project to open.
When you have finished, your kit should have:
one document or folder, with the review date at the top a section for each task type, with its prompt or guide, a when-to-use line and a watch-for line one page of data rules covering approved tools, allowed data, the redaction method and the recording notice time-saved results with their labels and quality notes the tasks you kept manual, each with its reason a note of who you shared it with and what they said
Start by gathering every piece you built into one place, before you tidy any of them.
Assemble the kit as one document or folder with every template, your data rules and your results, and share it with one colleague for feedback.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).