Andreas Holley

Andreas Holley

@duk.im

Backend Eng & Open Banking at Monzo Sometimes write stuff at blog.duk.im

54 followers91 following69 posts

Just had a bit of an awakening moment with agents, going through fly.io/blog/everyon.... Added an ls and cat tool using the same pattern as ping. Got the agent to inspect its own source code, it suggested adding write_file, and now the agent is modifying and improving itself 🤯. All in ~200 lines TS

Of course a lot of it comes from the model, using GPT-5.4 here, but even much smaller models manage the tool calls. Also it has no system prompt, perhaps aside from whatever’s baked into the model itself, so start of chat behaviour feels odd, but incredibly quick to start responding

Yeah this new agent team mode with Opus 4.6 and the pro plan are not made for each other, blew past my 5 hour limit in 11 minutes.

I tried out codex for some personal stuff (have a month free of the plus plan), and it seems like I have easily 10x the limits with 5.3 codex vs Claude code with the pro plan (same price, and I’ve been paying monthly since November). The desktop app seems pretty nice too

Meanwhile codex with ~5x the tokens used, so if anything I'm getting ~25x the usage?? Something seems not right here. Claude code is usually _fine_ when I use just sonnet in a single instance without a bunch of sub agents.

I tried out codex for some personal stuff (have a month free of the plus plan), and it seems like I have easily 10x the limits with 5.3 codex vs Claude code with the pro plan (same price, and I’ve been paying monthly since November). The desktop app seems pretty nice too

We have Claude with ~uncapped spend at work so I’m pretty invested on that side of things and make good use of it. But codex also lets me use the plan with Opencode so another thing to look into (I used opencode with Minimax quite a bit a few weeks ago and it was pretty close to Claude with Sonnet

For context - unlike most stations, but similar to other terminals, platforms at Euston are often announced late causing a big rush to get to the platform and get a seat. Various sites show 'booked' platforms the train is likely to come in to, this takes it a step further showing incoming delays

Started looking into static site generators this week, ended up going super deep exploring with Claude and now I find myself comparing dynamically generated og:images for a blog that has no content (yet).

Andreas Holley reposted

Now THIS is what a customer-centric app is. Most banks keep sending out warning emails about impersonation and scams. Any app can implement this capability with a little coding + backend work (assuming their BE sensible enough): and yet Monzo is the first one I’ve heard do it.

You are up to date.