Build Your AI Doppelganger
An AI doppelganger turns one repeated workplace task into a reusable AI-supported working system. Listen to the episode, watch the configuration process, then use the guide to choose a narrow task, set clear instructions and keep human judgement in the loop.
Start building your AI doppelganger with the podcast.
Listen for the shift from repeated ad hoc prompting to a reusable configured workspace for one task you understand well.
Watch the AI doppelganger configuration process.
The process is simple: choose one recurring task, gather reference material, write the assistant instructions, test the output and define the checks you will always complete before using the result.
An AI doppelganger is not a copy of you. It is a reusable setup for one task.
The aim is to stop rebuilding the same context, rules and examples every time. Keep the assistant narrow, make the repeatable parts visible and keep judgement, checking and final responsibility with you.
Turn repeated prompting into a reusable AI doppelganger.
Episode 17 is about building an AI doppelganger for one recurring workplace task. It moves you beyond repeated ad hoc prompts by configuring a reusable assistant with the task context, reference material, output rules and review boundaries already built in.
The point is not to automate your whole role or remove professional judgement. The point is to capture the repeatable parts of one task so you can get more consistent results and spend more time on the decisions that still need you.
Why repeated prompting wastes effort
Many people keep re-entering the same context, formatting rules, tone preferences, source material and spelling expectations into new chats. That can work, but it makes the process fragile. It also makes your AI capability harder to explain, test or improve.
A configured assistant gives one recurring task a dedicated workspace. It can hold approved examples, reference material, instructions and limits so the setup does not need to be rebuilt every time.
The four parts of a useful AI doppelganger
A practical AI doppelganger does not need to be technically complex. It needs a clear task, useful reference material, repeatable instructions and visible limits.
Target task
Name the recurring piece of work and define the output the assistant should help produce.
Reference material
Store approved templates, examples, definitions and source material that show the expected standard.
Processing steps
Describe the steps the assistant should follow when reviewing, sorting, analysing or drafting new material.
Limits and checks
Define what the assistant must not assume, invent, recommend or decide, and where human review is required.
The simple configuration sequence
Start with one task you understand well enough to test and supervise properly. A narrow assistant is easier to trust than a broad assistant expected to handle everything.
- Select one recurring task Choose work that happens regularly, follows a reasonably clear sequence and produces a recognisable output.
- Capture the current method Write down the steps you already follow when you do the task well.
- Add reference material Include approved examples, templates, definitions or source material that the assistant can use safely.
- Write the output rules Set the required structure, tone, terminology and New Zealand or Australian spelling conventions.
- Test against real examples Use examples you already understand, compare the result with your normal process and refine the instructions.
- Document the human checks Record where you will verify facts, adjust tone, check context and approve the final output.
Configure your assistant using official guidance
When building a configured assistant in ChatGPT, use the official OpenAI guidance for creating and editing GPTs. It explains how to configure instructions, add knowledge, test outputs and refine the assistant before relying on it for repeated work.
What to avoid
Do not start with an AI doppelganger for your whole role. Do not use sensitive or high-risk material unless your organisation’s rules clearly allow it. Do not treat the assistant as a decision-maker. It can support repeatable processing, but you remain responsible for the work.
For Zero to AI guidance on safe examples, read the AI learning terms of use. For wider guidance on safe, transparent and responsible generative AI use, review the New Zealand Government responsible AI guidance.
You can also use the AI Learning Labs to turn other Zero to AI episodes into practical workplace exercises.
Configure one AI doppelganger for one recurring task.
Choose a task you repeat often. Write the task purpose, add one approved example, define the output format and list the checks you will complete before using the result.
Use this AI doppelganger prompt as a starting point.
This is not a form and nothing is saved on this page. Copy the prompt manually into your preferred AI tool if it is useful.
Act as a practical AI assistant configuration coach for an experienced professional. Help me design a narrow reusable AI doppelganger for one recurring workplace task. Ask me what task I want to support, what output I need, what approved examples or reference material the assistant should use, what steps it should follow, what tone and spelling rules matter, what it must not do, what checks I must complete, and how I should test it on real work before trusting the results.
Make one workflow easier to repeat.
Build one narrow assistant, test it carefully and keep improving the instructions as your work and judgement become clearer.