iiinova — where your digital self keeps evolving

Sure, but the opposite can also be true. You spend too long contemplating that you don’t get anything in front of users for feedback, etc.

It’s not really meant to imply anything. AI is a tool, and it’s up to you how you use it. Sloppy code and shipping unfinished features isn’t just a problem with AI, human devs have been doing that for a whole lot longer.

140

I think that’s a bit too reductive. Treating the fact that a system can do X as evidence of the system’s purpose stretches the premise well beyond usefulness.

100

Chris Banes @chrisbanes.me · 12/08/2026

Ignoring a (very good) tool because you can write lots of sloppy features feels a bit like ‘cutting your nose to spite your face’. As a counterpoint, nearly all of Haze’s Glass impl was written by GPT. I orchestrated the approach, code, tuning & testing, and I am personally happy with the result.

360

You can also try everything out in your web browser here 👇 chrisbanes.github.io/haze/latest/...

chrisbanes.github.io

Haze samples

030

Chris Banes @chrisbanes.me · 12/08/2026

🚀 Haze 2.0.0-alpha05 has been released! 🔎 Glass visual effect (finally) ⚙️ New Style-like API structure 💨 HazePerformanceMode toggle This should hopefully be the last alpha release chrisbanes.github.io/haze/

1191

Chris Banes @chrisbanes.me · 03/08/2026

Here's a little preview of the new glass effect coming to Haze 2.0 Should be in the next alpha release

0244

Chris Banes @chrisbanes.me · 22/07/2026

🛍️ Shopping isn’t a category I got Hermes Agent to match marketplace orders with my bank transactions, then categorise each item by what I actually bought. Same spending. A much clearer picture of where the money went. chrisbanes.me/posts/shoppi...

chrisbanes.me

Shopping Is Not a Category

How I used Hermes Agent to match marketplace order emails with Actual Budget transactions, split purchases into individual items, and reveal where the money actually went.

041

Chris Banes @chrisbanes.me · 26/05/2026

Hunting down Jetpack Compose recomposition issues manually is a huge time sink I’ve spent the past few weeks using @mmckenna.me’s Dejavu to simplify this: - Write Dejavu tests to establish a baseline - Tell the agent to hunt for issues and A/B fixes - Profit

3181

Chris Banes @chrisbanes.me · 12/05/2026

🤖 I've just open-sourced a new set of skills, covering Kotlin, Compose and Android development. Very early days, but it covers a lot of things which crop up in code review regularly. github.com/chrisbanes/s...

github.com

GitHub - chrisbanes/skills: Skills for Kotlin, Jetpack Compose, and Android development

Skills for Kotlin, Jetpack Compose, and Android development - chrisbanes/skills

0453

Chris Banes @chrisbanes.me · 02/05/2026

Thanks! Ideally, distribute your own.

Chris Banes @chrisbanes.me · 29/04/2026

💠 Haze 2.0-alpha01 is out! This is a big release, splitting Haze up into layers, with the goal of enabling pluggable visual effects. Let me know what you think! chrisbanes.me/posts/haze-2...

chrisbanes.me

Haze 2.0: A Pluggable Visual Effects Engine

Haze 2.0-alpha01 turns blur into a pluggable effect, splitting it into a separate module and opening the door to custom visual effects built on top of a shared rendering infrastructure.

2297

Chris Banes @chrisbanes.me · 30/03/2026

This is opposed to Gradle + KMP builds (with iOS) which takes 15 mins, at a _minimum_.

210

Chris Banes @chrisbanes.me · 30/03/2026

Starting hacking on an AI Agent orchestration tool (for myself). Decided to build it in Rust (because why not), and holy cow: the entire CI run takes ~40 seconds, and most of that is setting up the toolchain. github.com/chrisbanes/e...

github.com

feat: ACP agent client, orchestrator state machine, and multi-agent p… · chrisbanes/ensemble@930a0e7

Contribute to chrisbanes/ensemble development by creating an account on GitHub.

360

Reposted by Chris Banes

sinasamaki @sinasamaki.com · 21/01/2026

✨ ChromaDial is a new #ComposeMultiplatform library that enables you to create beautiful Dials for your #JetpackCompose app Its API is similar to the Slider, but instead works in degrees Just pass in and set the degree state #buildinpublic

2255

Chris Banes @chrisbanes.me · 22/01/2026

Do you have any (rough) numbers of how much these cost at runtime?

120

Chris Banes @chrisbanes.me · 17/01/2026

The $20 sub for Codex is really generous for the price. I wouldn’t be surprised if it’s close to the CC $100 plan in terms of allowance

Chris Banes @chrisbanes.me · 16/01/2026

GPT-5.2-Codex is listed at 272k but I’m not sure if that is actually true. I can’t find anything to confirm that anyway.

Chris Banes @chrisbanes.me · 16/01/2026

I like Copilot (it’s basically a more basic version of Cursor), but beware that a lot of the models have a reduced context window size. models.dev?search=copil...

models.dev

Models.dev — An open-source database of AI models

Models.dev is a comprehensive open-source database of AI model specifications, pricing, and features.

Chris Banes @chrisbanes.me · 15/01/2026

OpenAI mostly (it uses oAuth so you can use your CGPT sub), and Antigravity (mostly for the free Opus/Sonnet allowance).

120

Chris Banes @chrisbanes.me · 15/01/2026

After becoming infuriated by Codex (the tool) lately, I've switched to @mariozechner.at's `Pi` and it's like a breathe of fresh air 🧘 All of the stuff I actually use, it's super fast, and most importantly it doesn't constantly ask permission to run `git` shittycodingagent.ai

shittycodingagent.ai

buildwithpi.ai

A terminal-based coding agent

2101

Chris Banes @chrisbanes.me · 14/01/2026

In some ways I really like GitLab, in other ways it’s infuriating. It’s CI system being top of my list…

020

Chris Banes @chrisbanes.me · 13/01/2026

+1 to this, although I personally prefer a mix of Codex 5.2 and Sonnet for coding (depending on task complexity). Codex is also my favourite for code reviews. Opus is too expensive to be a pure coder imo.

050

Chris Banes @chrisbanes.me · 29/12/2025

Definitely try other models. I know that Gemini 3 scores highly on benchmarks but I still find the Claude models the best in real world coding (by a good margin). I have a follow up blog post in draft which goes more into this.

250

Chris Banes @chrisbanes.me · 29/12/2025

Just get AI to review it? (Only half joking as I do actually do this)

Chris Banes @chrisbanes.me · 29/12/2025

Totally agreed, AI isn’t the only reason this happens.

Not sure. It doesn’t seem to run any builds anyway, as the commits are full of compile errors

just found the limitation of Copilot. Since it can't run builds/lints/etc as part of the Agent run, you end up doing a lot of bouncing around via CI. If your CI is slow (GitHub Actions are not exactly the fastest) then you get into a bit of loop.

100

Yep, but I did this solely from my phone and the GitHub app. The ecosystem moat that GH have is significant.

And that’s where we’re at. Still need to finalise a few things, but I’m sure that this PR will be landing later today without me touching a physical keyboard. 5/

270

Add ColorFilter support to HazeTint and refactor to sealed interface by Copilot · Pull Request #854 · chrisbanes/haze

Started an ‘Agent Chat’ via the GitHub app. I gave it the URL of the GitHub issue, and we iterated on the high-level proposed plan to fix. Once happy, I asked it to “Upload a Draft PR” It created a placeholder PR immediately: github.com/chrisbanes/h... 2/

github.com

Refactors HazeTint from a data class to a sealed interface with concrete variants for color and brush tints, adding ColorFilter support for advanced tint transformations (color matrices, lighting e...

120