Now, on the investment committee side I have tools such as replit working. It helped me to build a custom investment dashboard and also synthesise ideas. So we have a number of custom windows running that provide a really helpful level of data.
I had the idea of feeding a conversation I was having with Phil about our August podcast and…. so in it went!
We wanted to see if it could write a practical money book the way devs ship apps modular, versioned, and improved with feedback and ‘pull requests’. So we set up a live project and started vibe-coding chapters (structure, prompts, snippets, and edits) right in the open:
👉 Peek at the work-in-progress: BookFlow on Replit
What worked
Rapid drafts: outlines and “minimum viable chapters” came together fast.
Clear scaffolding: each chapter forces a decision + action + checklist (no fluff).
Easy to update: if rules or thresholds change, we version the chapter and keep moving.
Its not the best book you’ll ever read on finance, but it’s actually way better than I had expected (scott here writing todays column) so if you do have a look, let me know your thoughts!
What didn’t
Voice drift: AI + code = speed, but your voice changes if you’re not ruthless.
Compliance polish: examples must be tight and consistent with AFSL standards.
Editing rhythm: the fastest way to finish a chapter was still… a human editor.
The Super-Comparison Experiment (or: why free tools failed us)
On the pod this month, we tried to be laymen and compare supers using only free, public tools. It… didn’t go well. Here’s why:
Why the results were all over the shop
Different assumptions: tools model different return expectations, fees, and insurance by default.
Apples vs oranges: comparing your current option to a fund’s “flagship default” is misleading.
Fees hide in the footnotes: admin vs investment vs indirect costs many annualisers are inconsistent.
Timeframes cherry-pick winners: 1-, 3-, 5-year tables favour whichever style just had a run.
You ≠ the average member: insurance, salary sacrifice, spouse contributions and tax all shift outcomes.
What actually works (our process)
Get the real statement(s): current option, actual fees, insurance premiums, and contributions.
Normalise the data: calculate fees on your balance, not the brochure example.
Model like-for-like: same risk level, same contributions, same insurance, same time horizon.
Stress-test: show ranges (good/average/poor markets) rather than one neat number.
Decide with intent: lower fees are great but not if you lose benefits/insurance or take on the wrong risk.
As a Wealthlab client you get the benefit of all our tools already. But do you know somebody who’d benefit from our service? Send them this email and get them to book in a chat online!