Jamie Taylor

Then,

I'm trying to think of things that are abstract enough that probably apply to lots of different languages rather than just .NET things, because then only I understand it.

Yeah, yeah, exactly, yeah.

So the idea is you create a feature by defining a spec, then defining the plan,

and then splitting it into tasks.

If we go 10,000 foot view, you have a feature that you have.

Somebody comes to you and says, I want to check out feature.

That's your feature.

You create the spec for the feature.

It needs to use these kinds of, I'm expecting to be able to

click a button and input my credit card details, how they get there, it doesn't matter.

Input my address details, how they get there, it doesn't matter, but those need to be validated.

Then the plan is the how, not the what.

So you're saying in the plan, you're saying, so we're going to use Stripe,

And we're going to use this particular type of ORM for our Ruby app because we don't want to use SQL injection, you know, and then from the spec and the plan, and you've gone round in circles a few times with both of those steps to make sure that the

the coding agent, I'm trying to remain technology agnostic, right?

So then the coding agent is then pointed in the right direction, shoved in the right direction of, you're going to use these particular projects, these particular processes, these particular technologies.

Then you say, cool, you've got the plan, you've got the spec.

And the other thing it does is during the plan phase, it actually goes and does the research.

So what it will do is it will go, oh, cool, we're using Ruby version, I'm going to make up a number, 2.6.