halamadrid

On demand delivery where you pay a premium of 30-40% over the actual cost (sometimes even more) is definitely a luxury to me. I can probably afford to do this but I still do pick ups along with my other errands. Only is rare circumstances I get one off deliveries and it hurts me to see the cost each time I look at it!

Doing it regularly is probably wasting money especially in this economy.

Kids + full time - pick up the food on your way home? Medical condition - arrange for a routine cost effective delivery (like those meal delivery services who tend to deliver on their routes for free even?) work 2 jobs - pick up along the way somewhere can't drive? - use that routine delivery or find a way to cook or use a weekly meal delivery service etc.

But buying one off every other day paying the premium seems definitely a luxury. YMMV.

If there is enough demand, someone will build and supply it. So enough buyers have to buy the model and keep it running that long.

How complex is the claude code codebase itself? Assuming its primary responsibility is to orchestrate the user prompt with models? The risk of this is not as big as a model release itself right?

The need for actual lawyers will persist I think from my own experience. I attempted drafting a contract with some points myself using AI, but after several edits I wasn't sure if it was correct. Sending it to an actual lawyer ended up in so many corrections I couldn't imagine the first time. One big thing was the overly excessive protective clauses which didn't make sense for reality or conflicted with another.

Its just like code I suppose, if you can read and understand and validate, you can use it to scale and otherwise it could end up being a vibe effort.

If there is one company that I'd never ever use for running a personal agent, that would be Meta. But they have the advantage of a large user base who still uses the platform heavily and a working personal agent is a huge pull for most of their users.

The focus on tools like these is always on how you create. Which IMO is the trivial part. Folks will say you can use Claude to write the code. Claude can also write the definitions needed for tools like this. The big difference is always in how/where you’ll run it and how to monitor things and maintain it long term.

The runtime being the biggest value add from tools like this.

Like Airflow or Temporal or Unmeshed.

The value for me is the runtime not the ability to create stuff. Drag n drop is definitely dead. But so is coding by hand. Pick the right runtime for getting the value from tools like these.

I fail to understand the need to do this, isn't it obvious from the history that Apple will find out and will go after this? Why take this risk? I feel its better to start from scratch & redo things so that you get two things

1. a chance to correct problems you couldn't before because things evolved too much 2. don't break rules or agreements you have signed before

The first one is the key part - I think redoing something the second time almost always ends up in a better result. If you are paid to do it with enough resources and team then all the more easier.

What are all those oops for?

This is probably the first real thing that is affecting me personally with this whole AI business. Having to pay more for device upgrades going forward. I hope the demand settles or new memory production offsets the demand.