hakanderyal

To effectively use AI-as-an-engineer as a seasoned developer, you must change how you work. It won't work otherwise. Your knowledge & experience wouldn't allow you to ship raw AI code, you'll end up being frustrated instead.

An alternative way is using it as a pair programmer, which is one step above enhanced StackOverflow.

But they say "It comes with strong privacy and security built in" right on the page!

It's not, at least for now. Various staff members have confirmed this.

It’s optimized to death. Every word, every comma. I have a rule file for writing rule files.

Problem stems from using my own abstractions instead of common frameworks.

For a full stack session, backend + front end + docs about the part we are working on adds to that amount.

I’ve accepted it because it results in code exactly the style I would’ve written. It’s a good tradeoff to cut off the slop.

I have rule files that guides the agent towards my coding standards, code style, house rules etc. They alone cost 60-80k tokens, and they are the backbone of my system that prevents slop. Pre 1M context, I had to build complicated tooling to re-include the relevant docs to the context upon compaction, which relied on unstable transcription file format, which was a pain to maintain. With 1M context I deleted all of those. Nowadays most of my sessions uses 300-450k context.

Another thing that's preventing me from trying Codex. (the other is @ referencing files not auto including them to the context)

1M should be table stakes for frontier models at this point for programming.

UK, USA and Singapore are the popular choices for this. UK is simple and cheaper, Singapore is a bit more expensive. I don’t see a reason to go USA route unless you are seeking investment from there.

I’m using UK myself. It takes a few forms and half a day to get a working Wise/Stripe account.

I think this is a common sentiment among heavy users of AI that also still cares about code quality.

I've built up a skill harness and review flow that makes Opus generate slop-free code 90% of the time. But the remaining 10% requires me to stay at the helm. Especially in the early stages.

I would love to use loops to automate more, but I couldn't do it with the current generation models.

And on the back of my mind I'm still evaluating the possible future where we are forced to API pricing. I'm currently paying $400 for Opus, and use around 1.5-2 billion tokens per day. This will cost around $20k/m with API pricing. And I don't want to even imagine the possible scenario of getting locked out of frontier models because of politics.

Will the models get better to cut me out of the loop completely? I believe so. Will the open source models catch up tho SOTA models, and diversify from China-only? I hope so. Otherwise 2 superpowers will wield a soft power that can cripple the tech industries of all other countries.

CC can explain it clearly, which how I learned about how the inference stack works.

This is also the reason why we have two polar opposite views on AI. “Slop generator” vs “Next best thing since sliced bread”.

With SOTA models it all depends on how you drive them.

Even that would be more meaningful test. They basically coated the ball with a strong smell, then they prepped the dog with that smell, then set it loose in a 5x5 meter area.

"Our tests gave models the vulnerable function directly, often with contextual hints (e.g., "consider wraparound behavior")."

And it's very valid. Because otherwise you would ask Claude to trim a tree and it would go raze the whole forest and plant new seeds. This was the primary pain point last year, especially with Sonnet.

Similar usage here. But I encountered this moments, and I chalk it up to the random nature of LLMs. Back in Sonnet 3.5 days, it would happen every other day. I even build an 'you are absolutely right' tracker back then to measure it. Opus 4.6, maybe once or twice a month.

One added benefit is it allows you to throw more tokens to the problem. It’s the most impactful benefit even.

Context & how LLMs work requires this.

From my experience no frontier model produces bug free & error free code with the first pass, no matter how much planning you do beforehand.

With 3 tiers, you spend your token & context budget in full in 3 phases. Plan, implement, review.

If the feature is complex, multiple round of reviews, from scratch.

It works.

We are not there yet. While there are teams applying dark factory models to specific domains with self-reported success, it's yet to be proven, or generalizable enough to apply everywhere.