Worked on BEAR 2.0 the weekend.
The plan: combining Bookkeeping (incl. Doc Generation) + Client Portal + Projectmanagement + CRM into one.
Certainly possible, but a huge task.
Mainly did the designing in Figma.
Also came up with the „dynamic Angebot“ which can be accepted by button click instantly.
Had call with Leo about it, he said: can you inmplement till tuesday 9am – since we have a discovery call there. Said not sure.
I kind of dread the prompting.
Especially if I don’t have a pixel perfect design made in Figma.
I read recently that prompting is „demoralizing“ and I agree.
But on the other side, it opened HUGE opportunities.
A well crafted prompt and prepared context can build app screens in one go.
But architecting the design + data structure + logic -> then AI implements it.
This is the playbook.
Meaning I should go back to the DRAWING board and design all screens for the App out in Figma.
Gives me Figma practise.
But I started doing this Saturday or Friday and got frustrated (kindof) since I designed one screen I already kind of had and it took me 3 hours or so. (The Angebots Table – and I wrongly approached table design in figma)
I said to self: don’t rush design.
Otherwise I end up with half-assed shit.
My current dilema: I one-shot functionality and if I want to change things, I tell AI to rewrite the whole thing.
I have no maintainability, should we get larger codebases.
I read something about tools that can index a codebase smartly and make this possible though.
On that note: I have not yet tried Claude Code or Cursor for that matter.
I’ve been using lovable.dev to prototype UI’s, but mainly just Claude itself in the webapp to create html, css and js demos – as well as PHP code for wordpress.
Absolutely BLESSED to have AI -> but we have to make it count by learning how to DESIGN well in Figma, then giving it the design to implement. I believe this is the way.
I have some things in the current situation to ORDER: Open Client Projects, Taxes. NO TOOL Saves you here.
Do them with the google suite: Sheets, Docs & Drive is all the management tools you need to get those tasks DONE.
And I will. I’m executing the „Angebots Logic v0“ tomorrow as one-shot prompting.
But will design BEAR into a production grade app. -> be sure of that. Fits into the PB ecosystem: can sell it for 19-99$/mo. If we build it, we build it correctly.
Bruckner Mahler Studio will assemble a NextJS dev team.
We can still use WP as CMS headless. Depending on data structures and how much sense it makes to use ACF to write into custom fields then. (I know not much of that YET).
I want to do things right.
Build the empire.
Also use EXISTING TOOLS – if needed. If you can use google sheets, USE google sheets. Use tools if necessary. (for example: I barely use Paymo. – just use G Sheets).
Stop explaining that and do as needed. BRAIN: process this and make something of it.
Lessons: do what is necessary now (just like the CARP dream told me. Dont run away into „nice“ projects – something like BEAR is fucking complex and requires heavy brainpower, dedication and free brain capacity.
Build the agency ONE by ONE — at the RIGHT time.
End
Feeling fairly exhausted now, on a Sunday evening.
Call today with Leo: he’s heading to Innsbruck tomorrow to check a Wohnung that has 2 bedrooms + a living room spread on 2 floors. I could move in there.
That would be sicko. But we’ll see what they say and how it actually looks.
If there’d be room to run that.
But the fuckers would be at Zivi during the day, giving me the space.
It would be SICK if that worked out.
Thank you Lord.
Stoked about that.