Advancements Elsewhere - 19/09/2025

Four transfers later within the first month and I've reached £1,000 in cash ISA. It feels like progress. There is growing hope that I will reach my local target. I think it was around £12,500, I can't recall the exact end figure in the plan. I only remember what I need to do each week and stick to it. 

My second pension payments coming up and it might coincide with the transfer of a second lost Scottish Widows pension. I won't get tax relief on the transfer as it has already been paid but an £425 one off monthly installment will make a nice different to that plan.

I am still in the process of discovering lost pensions. I've missed one and it was an electronics assembler role I had when I was much much younger - a long time ago. I don't think it's great due to the rate I was earning at the time but I do think there should be some hidden cash somewhere as a result if that role for sure. 

GILT payments are in schedule. I'm not really making any money at the moment but that's okay. The Playbook Strategy does not call for returns during the fortress phase. So any pennies that I am seeing building are a bonus. My fourth payment will be approaching next Monday so I'm glad to see that I am so far committed to making this happen.

The Playbook and it's Companion Guide were becoming a worry. As I've been using ChatGPT to breathe life into this project I didn't fully explore its limitations.

I know it can get things wrong. What I didn't realise is that I could run out of space. So for several weeks ChatGPT is telling me it is crafting a masterpiece but I keep interacting with it, filling up its memory. It go to the point I had messages telling me I've maxed out the resources for that conversation. I thought I was going to lose everything that was in production and all of the backend data that was helping to crystalise the ideas into a material product. 

Within days I found a satisfact work around. In fact I asked ChatGPT in another conversation for a work around. It gave me options. It seemed to work, but it wasn't perfect. What was recovered from the other conversation came across patchy; it felt incomplete. So for a couple days I've been having to interrogate ChatGPT to recover the missing data. I get the impression there is reluctance. It locates a little bit and applies it as if to appease me out of pity. But I know that what has just been served up is not the full shebang. So I've found myself cycling the same questions over and over, resulting in being drip fed what I originally asked it to do. 

Thankfully, the Playbook Companion Guide is approaching a status where it is almost ready for final pass and polish, at least for this version.

I still think I am being short changed but once I have a decent master file for both Playbook and Companion Guide I can feed them both into another AI and ask it to find the flaws, gaps, weaknesses and have it suggest what would improve it. Then I can go back to ChatGPT with this findings and ask it to make the changes. Then I can do the same again or even ask another AI to run a simulation of the playbook content to see where it falls flat. 

I should be able to advance with this project more rapidly now I'm not being delayed by the AI having it tell me for months that it's working on section 7 when it probably wasn't.

At no other time has ChatGPT told me I am absolutely right than now. I don't really want it to keep doing that, I just want it to be right first time.

But I suppose, in using it so much I've been able to stress test limitations and find workarounds.

I thought I had to flip a toggle switch to have ChatGPT access all conversation that I've started. Turns out, if I did see a switch, it has gone and it does it any way. I found it odd that I mentions the toggle switch to it but it never revealed that it was already able to access all conversations without my say so.

So chatGPT is actually learning about me when I thought it wasn't. And I wonder if what it is learning has given it an attitude problem.

I guess I'll never know.