Follow for how to build ai era startups and products. Fun thing : https://xitter.cf/t.co/EMIolkwkt7 Main thing @devclocked. Prev: agency (exit) + £MM recommerce
London
Joined January 2026
- Tweets1.1K
- Following155
- Followers2.5K
- Likes203
Pinned Tweet
Everything I learned building a full game with GPT 5.6 Astra, start to finish. Long read.
Try one small test before starting another project: give the product with the clearest demand a 90-minute block with no redesigning, no feeds, and no new ideas.
If focus comes back once the choice disappears, too many open loops may be draining you.
Traction can make a project less fun. Building turns into support, decisions, expectations, and a growing list of work you did not plan to do.
Before treating it as a focus problem, ask whether you dislike the work the product now requires.
for anyone looking to vibecode saas replacements..
before you build any internal tool, build a scorecard
Subscription spend is easy to see, but keeping it running later is not.: fixes, access requests, integrations, pinned on the person who still knows how it works.
A highly specific workflow with deep integration needs can justify building. An ordinary workflow, a short deadline, or no clear maintenance owner leans toward buying.
Cost of failure can change the answer. A cheap subscription is not cheap if errors are expensive.
I'm not sure that new builders can build distribution on X from zero in 2026. Without an audience or early proof that people care, you're competing with years of accumulated trust.
Is X still a realistic first channel, or mostly an amplifier now?
A product idea needs two answers: can you build it, and can you reach people who need it repeatedly? The second should shape the product early. Features people can share or discover solve a different problem than features hidden behind a login.
A product can have demand and still be the wrong thing to pursue.
Ask whether you want to spend the next six months handling the problems that come with its growth. If the answer is no, that is useful information. It is not a character flaw.
Aside from Shopify plus, klayvio is my biggest expense for my ecom store. £900 per month is wild
Who’s vibe coded and alternative
Mobile coding has changing the game.
I can't believe how much i've been getting done with just my iPhone.
Currently wire-framing out my orchestrator app before i buy the duo
The iPhone DUO could take us to a whole other level...
Has ai fundamentally challenged the view of “focus”
Can we finally build multiple things with success guilt free?
How is this ok - seems like cloning and copying is just fine now
If these guys are doing it everyone else is cooked
OK so Opus 5.5 SMOKED GPT Astra at 3d modelling this stadium from old plans and reference pics
This was a near one shot. 1000x better than Astras one shot (check reply for Astra’s one shot)
I gave them the same prompt and the exact same reference plans 👇🏻
This was Astras version, and this was after 4 rounds of feedback.
This being said, I’m having great difficulty “finessing” the stadium and the surrounding areas.
I feel like I’ve hit max cap with opus right now and every targeted fix of the streets and certain things seems to create a knock on effect
Could be a skill issue
Astra will probably do a better job of finishing it.
We shall see
