Understand. Leverage. Stay Ahead.

Score Yourself
Against AI

Compare your writing strengths with a frontier AI model across five key dimensions. Discover where AI excels, where you lead, and how to work together more effectively.

Start the Assessment →

Takes less than 5 minutes

Score Yourself Against AI

You Frontier AI

About this tool

What is this?

A quick self-assessment on AI, through 5 simple questions.

How does it work?

You score yourself across 5 key dimensions. Your inputs are compared to a frontier LLM baseline. AI tools don't state what their capabilities are—but understanding that is the secret to getting the most out of your prompts.

Who's it for?

Senior finance executives or anyone keen to assess themselves against a frontier model.

What do you get out of it?

A visual idea of where to lean on AI and where you play a part.

Compare the outlines

The same brief. Two different approaches.

Below are the outlines for the same article. On the left is Claude's version generated from a prompt. On the right is the custom outline written specifically for finance executives.

CLAUDE

Title – Using AI to write better in finance while staying credible and human

Synopsis – A working banker's method for writing with AI without sounding like it. It covers where the model helps, where it does not, and how to keep your voice and your numbers right. The core is a routine you can run on a memo, an IC paper, or a client email.

Ideal length – 1,800 to 2,500 words

Outline –

  1. The credibility problem. Why AI prose gets caught and what it costs.
  2. What AI does well. The low-risk work it speeds up.
  3. What to keep for yourself. The judgment and numbers a model cannot supply.
  4. Keeping your voice. Draft the substance, then let the model edit.
  5. The accuracy rule. Check every figure the model gives you.
  6. Confidentiality limits. What you never paste in.
  7. A working method. The routine for a memo, IC paper, or client email.
  8. Editing out the tells. The marks of machine prose and how to cut them.

MY OUTLINE

Title: How I supervise AI to automate business writing without damaging my credibility

Synopsis: a practical guide on how to create long-form documents such as articles, credit memos and board papers more efficiently using AI. The deep dive includes examples, methods and frameworks that I regularly use to ensure documents are accurate, retain my distinct identity, and are far less likely to be binned as "AI slop."

Ideal length: ~2,500 words deep-dive

Outline:

  1. What are you trying to achieve with your writing?
  2. How to identify which types of writing can be fully automated?
  3. Overview of strengths and weaknesses of AI
  4. How to work with AI to co-author high stakes documents
  5. How to set up Claude to create long form documents aligning with your "house style."
  6. Conclusion: AI when set up and supervised correctly produces high quality drafts you can refine and personalise

Why the custom outline beats the AI version

The AI outline is not bad. It is well organised, sensibly ordered, and covers the topic. But asked to produce a competent structure, the model passed; asked to produce this piece, it could not. Here is the difference, point by point.

AI Outline
Custom Outline
What it's about
Treats the topic as a balanced how-to, eight even sections
Decides the commercial point comes first, that senior readers bin AI writing, and builds everything on it
Frameworks
Uses the expected shape of an article
Brings a stakes-versus-customisation matrix and a four-stage writing model the AI version doesn't contain
Evidence
General claims about what AI does
Builds in an experiment that tests its own argument, the comparison you're reading
Position
Even-handed throughout, takes no side
Ends on a verdict about what AI can and cannot be trusted to do
Currency
Works from patterns in older text
Uses recent, named cases the model can't reach for
Audience
Generic professional reader
Written for a specific finance executive and the documents they handle

The pattern is consistent. The model is strong in the middle of the work: gathering, structuring, drafting from what it is given. It is weak at the two ends: deciding what to say, and deciding what it means. Both of those came from knowing the topic and the reader, which is the one thing the model has no access to. That is why setting direction stays with you.