Strategic Insight
- Laurence
- Apr 7, 2024
- 3 min read
Updated: May 19, 2024
Link to the tool: https://chat.openai.com/g/g-GXDUnBAPh-strategic-insight
Contents
1. Aim
Best suited for
In-house change leads / teams.
Strategy / management consultants.
Tool summary
GPT Name: Strategic Insight
A tool to identify, structure, and draft any type of change document for any type of organisation. Examples include:
a diagnostic report into the current state of an organisation, business, or function.
a roadmap for your existing change plans.
any form of technology change specific documentation.
Use cases
Identify - relevant change documents.
Create - structure and create relevant change documents for any and every part of an organisation’s change lifecycle.
Delegate & review - Structured workflow to ease delegation to team and easier review of how the AI output was generated.
2. Workflow
The tool follows a simple 4-step workflow.
The tool is designed as a co-pilot, which requires frequent user review, expertise, and intervention.
If we include a rough split of activity between the AI tool and human user, it would look like the below.
To make this more real, I've included an example walkthrough of the tool below, which follows this 4-step workflow.
3. Example walkthrough
Step 1: Tailor context
The tool asks the user to consider a few questions to help tailor their output.
Note - The questions in this step can be partly or fully skipped.
Step 2: Select output type
Once the input questions are answered or skipped, the tool first summarises your answers...
...and then provides some options for the user choosing their output.
This includes a choice between asking for your own output, or answering some multiple-choice questions (e.g. 1b, 2a, etc.) to define the output.
Step 3: Create output structure
In the example below the user chose to answer the output multiple choice to define the output.
Then, the tool creates a skeleton structure of the output (which the user can then review and either keep, change, or discard as they see fit).
In the example below, the user chooses to amend parts of the outlined structure.
Step 4: Create output content
The user, now happy with the outline, can ask for an initial draft.
Note - if the user had any preference on tonality, writing style, etc., this could be a good place to include this request.
Additional features
This wraps up the main workflow. However, some users will need to create their content whilst juggling meetings or other interruptions that disturb their thought process.
To help with this, the tool includes:
A. Help commands
B. Reference data
A. Help commands
A help menu, to jog the user's memory about where they left off with examples of tool optionality.
This includes options to:
change the user's inputs,
change the defined output,
review continuation options
review reference data (i.e. change or company dimensionality)
By way of example below, the user accesses this menu using the slash command "/help".
B. Reference data
Some help menu options are for reference data.
The aim of these options is to serve as a completeness/sense check for the user when creating content.
By example below, the user accesses reference data for company segmentation by using the slash command "/company".
This sums up the walked example.
If seeking to build your own tool, you can see the main design principles and core features that guided the thinking behind this tool below.
4. Design principles
Existing principles
Mirrors all design principles on the 'About' page.
Additional principles
Guide the user through a simple workflow to:
quickly specialise to their use case.
ensure completeness in their own prompts and thought process.
5. Core features
Targeted focus
Filtering questions - to help the user (1) quickly narrow their focus to get the most relevant results and (2) capture all the user’s key requirements (a typical user’s prompts do not include all the information needed).
Risk mitigation
Audit trail – summaries of inputs to (1) help quickly identify and correct user or AI error or misassumption, and (2) help teams to review the high-level steps used to generate the AI output.
Lists of relevant dimensionality – available as reference lists or checklists to give the user comfort that all their considerations have been included.
Improved outputs
Chain of thought (CoT) prompts – use of proven CoT techniques to improve outcomes.
Output wireframe – to help the user identify and/or better amend the structure to their requirements.
Ease of use
Visible workflow – allowing the user to interrupt and “course correct” the AI generation process, as their thought process develops.
Optional & skippable steps – to allow the user to jump in at the relevant stage of the workflow and/or skip irrelevant steps.
Easy continuation – optional selections (e.g. A, B or C) to reduce user effort for thinking and typing (for time-poor users and/or users with frequent task switching).
Help menu – with shortcut slash commands.
You can access the tool at the link below.
