How I created a Resilience Maturity Model from Scratch (with AI)

And How I Had AI Build a Clickable Tool

I love playing around with AI, and my latest "toy" is Claude Sonnet 3.5. It's fantastic for programming and math. Recently, I created a comprehensive Resilience Maturity Assessment from scratch in just 15 minutes. Here's what I did:

1. Searched for a Research Paper on Resilience Maturity

2. Uploaded it to Claude Sonnet 3.5

3. Asked the AI to Read it in Detail

4. Applied My Custom Prompt Template

5. Visualized the Results

This process is not only easy and fun but also incredibly impressive. The tool I developed provides clear, actionable insights into organizational resilience, helping teams quickly identify areas for improvement. By automating this assessment, I’ve made it accessible and efficient for anyone to use, regardless of their technical background. Plus, you gain valuable knowledge about resilience frameworks and AI capabilities along the way.

Here is what I did:

Step 1: I downloaded the scientific paper from Vogus & Sutcliffe from Research Gate. The title of the paper was Organizational Resilience: Towards a Theory and Research Agenda. This is not a particular Resilience Maturity paper, but it worked anyway. Try with the paper, book, or document you like.

Step 2: I uploaded it to Claude Sonnet 3.5 and prompted:

Act as a Resilience Expert with over 25 years of experience. You focus on Resilience Maturity Assessments. Read the attached paper in full detail. Become the top expert in this topic. Then, summarize it for me and create a resilience maturity assessment for me. The Resilience Maturity Assessment must be usable as a stand-alone version (e.g., downloadable HTML). Create an interactive assessment, ask first the questions, and then create an interactive dashboard to visualize the results. Create max. 10 questions. Take a deep breath and start step by step.

Note, the prompt wasn’t very advanced nor done with love ;).

It still worked.

Step 3: Get the results.

The first result was not good. The questions were not good, and the visualization was off:

Then the cumbersome work starts.

I prompted several times to “analyze what went wrong,” “fix visualization,” or “put visualization on a second page,” and other prompts.

Be aware, it will take some time. The problem with Claude AI is that you run out of prompt credits relatively quickly (compared to ChatGPT). That means you have to wait for a few hours until you can proceed. But at some point, it will work out.

And keep in mind, this is only the “programming part” which can be tricky.

The whole framework, questions, rating suggestions, and instructions are done in seconds by AI, which already is of great help.

Of course, we want more.

We want the AI to program us a clickable tool which we could share, creating a certain “wow” effect.

In the end, I succeeded.

This HTML-based version offers 10 questions. Somewhere between 15 and 20 questions, it got too complex for .html, and I constantly ran into errors.

Of course, Claude could use another programming language with more capabilities. But then it is more complicated to run the tool (at least for a layman).

Here is the final result. In the first picture, you see the questions:

In the following picture, the beautiful (automated) visualization:

Unfortunately, I can’t attach the .html file in the email.

But what I can do is copy and paste the whole .html code so you just need to copy it, paste it in e.g., notepad, and save it as .html.

Next step, just double click it and test it. You’ll find the code below (save some cumbersome ours and copy and paste it ;):