Yes.

You deny it see, but.. ah bugger.. I hadn't planned for this outcome..

How can you "lose steam" on an admission of one of your own arguments?

Didn't help connecting it to my Macbook, but still..

We had super bursty traffic, and had to go with Google Cloud (very early days! [0]) because you'd need to communicate with AWS and pre-warm the ELB capacity of your expected bursts.

We did a dead launch to 60 million customers (0 to 60 million, no organic growth phase) this way. I wouldn't want to do that on a VPS.

[0] https://cloudplatform.googleblog.com/2013/11/?m=1

DX is simple, integrations between the two, and the stack is well understood by the LLM.

Lovable uses supabase, and is surprisingly easy to eject from too; I've done the lovable to Vercel + supabase a couple of times, even managing to keep it syncing via the Git integration. You can get proper scalable infra and minimal vendor lock in whilst the vibe coder gets to play with the pretty.

Still think they've not been matched for ease of "start a session, walk away, carry on somewhere else" as if you've never left your desk.

Muscle memory is a bitch!

And yet.. building shit is no longer the sole domain of the software engineer.

That's the sea change.

I've literally had finance and GTM stand things up for themselves in the last few weeks. A few tweaks (obviously around security and access), and they are good to go.

They've gone from wrangling spreadsheets to smooth automated workflows that allow them to work at a higher level in a matter of months.

That's what all this AI is doing. The shit we could never get the time to get around to doing.

Anything to invigorate the desktop.

Microsoft buying OpenAI.. 10 minutes later it's rebranded Copilot.. and.. nothing much changes in the world. Oh, except all the AI improvements are around Enterprise governance.

Now count the Amazon deliveries in a year on said same street. And next year, and the year after, and.. however long one keeps a Ford these days..

It's quite a scary thought exercise.

Whist I can't usually start from the exact same point in the decisioning, I can usually bootstrap a new session. It's not all ephemeral.

To your edit: I find that the most galling thing about finding out about the thinking being discarded at cache clear. Reconstruction of the logical route it took to get to the end state is just not the same as the step by step process it took in the first place, which again I feel counters your "feelies".

print(X) is a great example. That's going to print X. Every time.

Agent.print(x) is pretty likely to print X every time. But hey, who knows, maybe it's having an off day.

Are you seriously saying that breaking a large complex problem down into it's constituent steps, and then trying to solve each one of them as an individual problem is just a sensation of rigour?

Yes, it's quicker to do it yourself this time, but if we build out the artifacts to do a good enough job this time, next time it'll have all the context it needs to take a good shot at it, and if you get overtaken by AI in the meantime you've got an insane head start.

Which side of history are you betting on?

Get it to write a context capsule of everything we've discussed.

Chuck that in another model and chat around it, flesh out the missing context from the capsule. Do that a couple of times.

Now I have an artifact I can use to one-shot a hell of a lot of things.

This is amazing for 0-1.

For brown field development, add in a step to verify against the current code base, capture the gotchas and bounds, and again I've got something an agent has a damn good chance of one-shotting.

When pushed it did the 'ol "whoopsie, silly me"; turned out the hallucination had been flagged by the agent and ignored by Opus.

Makes it hard to trust it, which sucks as it's a heavy part of my workflow.

Self-starters. Idioms indeed!

At least that's the reason I've been given every time I've tried to take a contractor permanent!