Episode 17: Build Your AI Doppelganger

Episode 17 Learning Lab

Build Your AI Doppelganger

Stop rebuilding the same instructions every morning. Create a configured assistant for one recurring workplace task, with your reference material, output structure and New Zealand or Australian language rules built in.

Start the Lab Build your assistant Switch to Classic View
Before you choose your pathway

Turn a repeated workflow into a reusable working system.

This exercise is not about collecting another tool. It helps you document one useful workflow, make the results more consistent and create visible evidence of how you work.

What this lesson is about

This Lab addresses the repeated setup work hidden inside everyday prompting. You will isolate one frequent, text-heavy task and configure a dedicated assistant around it.

The aim is to capture the repeatable parts of your method while keeping judgement, checking and final responsibility with you.

By the end of this Lab you will be able to

  • ✓Identify repetitive workplace tasks that are suitable for structured AI support.
  • ✓Write clear assistant instructions that preserve New Zealand and Australian spelling, terminology and output standards.
  • ✓Define the points where AI-supported processing stops and human checking begins.
Key concepts from this episode

The foundations of a useful configured assistant.

The manual prompting loop

Repeatedly re-entering context, formatting rules and writing preferences into a new chat.

The reusable workspace

Using a dedicated project space to keep approved examples, templates and reference material together.

Consistent language standards

Building New Zealand and Australian spelling, terminology and formatting expectations into the assistant instructions.

The human checkpoint rule

Documenting where the assistant must stop and where human review, judgement and approval are required.

Choose your working depth

Choose the amount of work that fits your week.

Listen

Start with the episode.

Listen for the shift from repeated ad hoc prompting to a reusable configured workspace within the Learning Lab approach.

Watch

See the configuration process in practice.

Watch the worked example showing how to define task rules, organise reference material and set a clear output structure.

Read

Set the boundaries before you build.

Review the task filters and setup sequence before you configure your assistant.

Task selection filters

  • 1High frequency: Work completed weekly or fortnightly.
  • 2Rule-based: Work guided by a clear, repeatable sequence.
  • 3Text-heavy: Tasks centred on drafting, summarising, sorting or analysing text.

The configuration cycle

  • 1Isolate the task: Identify a recurring workflow that consumes meaningful time.
  • 2Capture the rules: Document the steps and standards you already use.
  • 3Define the limits: State what the assistant must not assume or invent.
Apply

Configure your assistant.

Capture the task, references, instructions and limits that will form your assistant playbook.

Required

Choose one recurring text or data task with clear steps and a defined output.

Required

List the approved references, definitions, examples or templates the assistant should use.

Required

Describe the step-by-step sequence the assistant should follow when analysing new material.

Standard / Ambitious

Specify the headings, tone, terminology and New Zealand or Australian spelling rules to follow.

Standard / Ambitious

State what the assistant must not do, assume or decide.

Ambitious

For the Ambitious pathway, identify a second recurring task for a separate assistant.

Copyable assistant instruction prompt

Act as a specialised workspace assistant configured only for the [Insert Task Name] workflow. Follow these instructions: 1. Review the supplied material against the approved reference files stored in this workspace. 2. Follow this sequence: check the input structure, identify the required deliverables, and map the relevant information to the agreed categories or internal codes. 3. Use New Zealand and Australian spelling and terminology, including words such as prioritised, organised and analysed. 4. Present the result under clear H2 headings. When information is missing or ambiguous, stop and ask for clarification. Do not invent details or make unsupported assumptions.

Reflect

Define your human checking points.

The assistant can handle repeatable processing, but you remain responsible for checking the work and approving the final result.

Required

List the non-negotiable checks you will complete before the output is used or shared.

Standard / Ambitious

How will you use the recovered time for higher-value work, learning or collaboration?

Create

Create your Custom Assistant Playbook.

Complete the required sections to produce a practical record of your assistant configuration.

Required

Summarise what this configured assistant demonstrates about your repeatable AI-supported capability.

Custom Assistant Playbook

Generated from your Episode 17 Learning Lab responses.

What happens next

Test your configured assistant on real work.

Save the playbook somewhere you can reuse it. Test the assistant on a real task, record what changed and keep the evidence for future reviews, proposals or portfolio examples.

Return to your assistant playbook
Saved
' + ''; }fields.forEach(function (field) { field.addEventListener('input', function () { saveField(field); renderAsset(); updateProgress(); showToast('Saved'); });field.addEventListener('change', function () { saveField(field); renderAsset(); updateProgress(); }); });pathwayInputs.forEach(function (input) { input.addEventListener('change', function () { setPathway(input.value, true); }); });pathwayButtons.forEach(function (button) { button.addEventListener('click', function () { setPathway(button.getAttribute('data-pathway-choice'), true); }); });const copyPromptButton = lab.querySelector('#z2ai-copy-prompt'); const promptText = lab.querySelector('#z2ai-prompt-text');if (copyPromptButton && promptText) { copyPromptButton.addEventListener('click', function () { copyText(promptText.textContent.trim(), 'Prompt copied', 'Copy failed', showToast); }); }const copyAssetButton = lab.querySelector('#z2ai-copy-asset');if (copyAssetButton) { copyAssetButton.addEventListener('click', function () { copyText(getAssetText(), 'Asset copied', 'Copy failed', showToast); }); }const printAssetButton = lab.querySelector('#z2ai-print-asset');if (printAssetButton) { printAssetButton.addEventListener('click', function (event) { event.preventDefault();renderAsset();const printWindow = window.open('', '_blank', 'width=900,height=1100');if (!printWindow) { showToast('Print window blocked'); return; }printWindow.document.open(); printWindow.document.write(buildPrintHtml()); printWindow.document.close(); printWindow.focus();window.setTimeout(function () { printWindow.print(); }, 350); }); }const clearLabButton = lab.querySelector('#z2ai-clear-lab');if (clearLabButton) { clearLabButton.addEventListener('click', function () { const confirmed = window.confirm('Clear the saved responses for this Learning Lab in this browser?'); if (!confirmed) return;window.localStorage.removeItem(storageKey);fields.forEach(function (field) { field.value = ''; });setPathway('minimum', false); renderAsset(); updateProgress(); showToast('Responses cleared'); }); }restoreState(); }labs.forEach(initLab); })();