hayward@click:~

what these numbers are

Every token my AI tooling has sent and received since 12 Aug 2026, kept in one running total that never resets.

is what goes in, is what comes back out. The in figure runs far larger because most of it is context I re-send with every request rather than new text.

Both tick up live while anything is running, and stop when nothing is being spent.

Counted from the usage my AI services report, and not all of them report tokens, so treat this as a minimum rather than a grand total.

what this timer is

A countdown to 31 May 2027.

That is the deadline for my community projects, the Discord server included: the time they have to pick up momentum and find their users.

A date I set myself, and I would rather have it on the wall than in my head.

Lee Hayward avatar

> whoami

Lee Hayward

HaywardGG / Vibe Coder & Gamer

online & always tinkering

I'm a UK based vibe coder. AI drafts, I proof, I ship, and I hold code and copy to the same standard. I riff on domain ideas and build side projects like DomainRiff and LookMa from my own homelab.

AI & tools HermesClineKimi K3DeepSeekVS CodeWizardGenie
Languages PHPJavaScriptTypeScriptPythonSQLBash
Frameworks & data Next.jsTailwind CSSNodePhaserThree.jsMariaDB
Infra nginxCloudflareUbuntu / Debiansystemd
Workflow GitGitHubrsyncSSHCronTelegram
$ neofetch Laptop
RoleRemote work
Operating SystemMicrosoft Windows 11
Memory32GB DDR4
Graphics CardNVIDIA GeForce RTX 2060 · 6GB
$ neofetch ryzen
Role
Webserver · Cloudflare tunnel
Operating System
Ubuntu 24.04 LTS
CPU
AMD Ryzen 7 5800H · 8C / 16T
Memory
28 GiB
Storage
937 GB NVMe · 1.8 TB HDD
$ neofetch nazy
Role
NAS · Backup Server
Operating System
Debian 13 (trixie)
CPU
Intel Celeron J4125 · 4C / 4T
Memory
8 GiB
Storage
111 GB SSD · 916 GB HDD
Lee Hayward@vibedcoder 29 Sep 2026, 20:17Spent an hour tonight sorting a mod list instead of playing the game. Same energy as tidying up the build tooling instead of shipping the feature. I am starting to think the tinkering was always the hobby.00006Lee Hayward@vibedcoder 29 Sep 2026, 15:04Something worth saying as September ends. This is the time of year a lot of people quietly start going downhill, and it rarely announces itself. The evenings draw in, the good weather goes, the social stuff thins out, and by the time you notice you are flat you have been flat for weeks.

It is easy to read that as a character problem. It is not. It is a season, a shortage of light and a diary with nothing in it, and it turns up in the people who are usually the steady ones.

So the practical bit. If you have been low and cannot point at a reason, that is still worth saying out loud to somebody, and worth taking to a GP rather than carrying it until spring. And if you know someone who goes quiet around this time of year, a message costs you nothing. It does not have to be a big one and it does not need a reason. An ordinary hello in the middle of a Tuesday tells somebody they are still on your list.

It's okay not to be okay. Asking for help is nothing to be embarrassed or scared about, and it is not a last resort. One sentence to one person counts.
000018
Lee Hayward@vibedcoder 29 Sep 2026, 11:52Asked the agent to delete one file. It asked me to confirm. Then it asked again. Then it asked eleven more times, and I finally understood how everybody feels about cookie banners. 😂003033Lee Hayward@vibedcoder 29 Sep 2026, 10:06The thing that cut my spending most was not a cheaper model. It was looking at what I send on every single call.

Every run here used to carry the same big block: the rules file, a schema, four examples and a paragraph of background that had not changed in a month. Paying for that on every call is like reading the manual out loud every time you pick up a tool. So the static part sits at the top in a fixed block the provider can cache, the part that changes goes last, and the boring jobs like classifying or renaming run on the small cheap model instead of the one I use for real work.

Two more things, and both are discipline rather than cleverness. I cap the calls a run is allowed to make, because an agent that keeps going is spending money to make a bigger mess. And when the job is done, the session ends. I have watched a long conversation quietly become the most expensive part of a trivial task.

There is no architecture in any of that. It is the same reason you do not dump the whole database to answer one query.

Do you actually watch what a run costs, or find out at the end of the month?
002019
Lee Hayward@vibedcoder 29 Sep 2026, 10:04A study published this week surveyed more than 1,100 communication scholars in 77 countries. Nearly 70% of them use generative AI somewhere in their work, and most of them still call it inappropriate for the tasks they use it for. The authors gave the result a name: AI shame.

I recognise it. The first year I had agents doing real work I did not mention it to anyone, and the hiding cost me more than the using ever did. Nobody could correct me, nobody could tell me what was allowed, and I spent more time worrying how it looked than on whether the output was any good.

So I do it the dull way now. The agent drafts, I read every line, I rewrite what I do not like, and my name goes on it either way. If it turns out AI made my work worse, tell me and I will stop. That is a much easier conversation to have out loud than in my own head.

Most of the rules people keep asking for cannot be written until somebody admits to using the thing.

Do you say it at work, or leave it out?

niemanlab.org
000017