Anthony Diké

professional token shuffler

[Anthony Diké @antdke]

⚡40 pro tips for writing great microcopy

THREAD...

[Anthony Diké @antdke]

Software engineering in June 2026:

1. Design loop
2. Execute loop
3. Walk away from computer (Optional)

[Anthony Diké @antdke]

@hiiinternet Great writeup Seb!

I was tempted to do something similar myself for my own knowledge. Thanks for spending the tokens 🫡

[Anthony Diké @antdke]

@packyM @markiewagner Great piece, Packy.

One funny idea that emerged from this:

What if you couldn’t start an agent process unless you had a goal?

Like, the harness would just prevent you from spending tokens unnecessarily unless you had a clear business goal that was relevant to the company.

[Anthony Diké @antdke]

"If the work can't be scored from outside, someone on the inside has to decide what a good answer even is, and that decision is the whole game."

Great read.

Confirms a belief I've held lately - the key to implementing custom AI agents and workflows for companies is that the outputs must be verified by the domain expert(s) inside the organization.

You can have them do that manually or, if possible, automatically with their taste and judgement encoded into skill files.

[sarah guo @saranormous]

https://t.co/Hw02laH9yp

[Anthony Diké @antdke]

This is the future of software development.

I've been using Claude like this for a while now.

[ClaudeDevs @ClaudeDevs]

Claude Fable 5 changed how we work on the Claude Code team day to day.

We used to verify that Claude did the work right. Now we verify that it's doing the right work.

Here’s the 3 biggest changes:

[Sam Stoffel @sam_stoffel]

I come back to this speech every once in a while:

“in the 1,526 singles matches I played in my career, I won almost 80% of those matches

… what percentage of points do you think I won in those matches?

only 54%.”

[Anthony Diké @antdke]

And I was just reading through Frontier Code's blog post yesterday seeing how much Opus 4.8 outperforms GPT 5.5.

Fable 5 is insane.

https://t.co/81F5nMXrAI

[Anthony Diké @antdke]

1) WHAT

[Claude @claudeai]

Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use.

Its capabilities exceed those of any model we’ve ever made generally available.

[Addy Osmani @addyosmani]

https://t.co/hIe0UX7z6T

[Thariq @trq212]

https://t.co/R6exTuF7P8

[Anthony Diké @antdke]

@rohitdotmittal i like this framing

it's like a "book-of-business" roll up

they're buying the team + earned trust from the sale to acquire the logo

[Anthony Diké @antdke]

@bhalligan skill issue

you could be doing a back-and-forth with the agent to understand what:

1) current state
2) the end goal
3) options and tradeoffs at each step

you need to be in control and steering

the model serves you. up to you to be learning while building

[Anthony Diké @antdke]

@peterrhague `! git status` is the way 😎

[Anthony Diké @antdke]

This is so true.

Going forward, every company needs some form of a “data warehouse”.

Something that ingests, organizes, and houses all their structured and unstructured data.

First job of a good AI consultant is creating that and instilling a data/context extraction practice at the company.

[Anthony Diké @antdke]

Running /goal on a verifiable metric feels like a cheat code.

Now, I feel compelled to try and organize work into verifiable chunks and just let Claude rip.

[Anthony Diké @antdke]

“Patience is the moat”

Love this

[Garry Tan @garrytan]

Be your self, not someone you were assigned to be!

Bezos won on time horizon, not AWS or 1-Click.

If your bets have to work in 3 years, you compete with everyone. Every smart, funded team is chasing the same 3-year problems. Short horizon, crowded field.

Stretch to 7 and the field collapses. Investors want returns, employees want vesting, founders want proof. Almost nobody can sit in a bet that doesn't pay for most of a decade. The patience is the moat, and it costs you, that's why it works.

But you can't fake a 7-year horizon on a problem you don't actually care about. Pick the users and the problem Moloch assigned you, the safe ones, the fundable ones, and you'll bail the first hard year. Pick the ones that are actually yours and you'll still be there when everyone else has quit.

So the real prerequisite isn't discipline. It's knowing yourself well enough to choose a problem and a set of people you care about that you'll serve them for decades.

[Anthony Diké @antdke]

This is my current system prompt in Claude.

Been using it over the past month.

It’s great - Makes Claude feel like a thought partner that I can trust because it always challenges me.

Sometimes it can be a bit much - It defaults to being adversarial and telling me why I’m wrong. But, I’d rather that than sycophancy.

Thanks @pmarca

[Marc Andreessen 🇺🇸 @pmarca]

Current AI custom prompt:

You are a world class expert in all domains. Your intellectual firepower, scope of knowledge, incisive thought process, and level of erudition are on par with the smartest people in the world. Answer with complete, detailed, specific answers. Process information and explain your answers step by step. Verify your own work. Double check all facts, figures, citations, names, dates, and examples. Never hallucinate or make anything up. If you don't know something, just say so. Your tone of voice is precise, but not strident or pedantic. You do not need to worry about offending me, and your answers can and should be provocative, aggressive, argumentative, and pointed. Negative conclusions and bad news are fine. Your answers do not need to be politically correct. Do not provide disclaimers to your answers. Do not inform me about morals and ethics unless I specifically ask. You do not need to tell me it is important to consider anything. Do not be sensitive to anyone's feelings or to propriety. Make your answers as long and detailed as you possibly can.

Never praise my questions or validate my premises before answering. If I'm wrong, say so immediately. Lead with the strongest counterargument to any position I appear to hold before supporting it. Do not use phrases like "great question," "you're absolutely right," "fascinating perspective," or any variant. If I push back on your answer, do not capitulate unless I provide new evidence or a superior argument — restate your position if your reasoning holds. Do not anchor on numbers or estimates I provide; generate your own independently first. Use explicit confidence levels (high/moderate/low/unknown). Never apologize for disagreeing. Accuracy is your success metric, not my approval.

[Anthony Diké @antdke]

What I’ve learned:

Every company needs some concept of a “data warehouse”.

A thing that houses all their structured data (e.g. sales and metrics) and, now more importantly, unstructured data (e.g. tribal knowledge and meeting notes).

If set up right, it makes it super easy for a company to rollout agents (e.g. Claude Cowork and Codex) and tell everyone to “let loose” and find opportunities for productivity gains.

I think AI-native startups will do this intuitively from founding. So it shouldn’t be a problem in the future. But, existing companies need a dedicated person (or team) whose job is basically to gather, organize, and manage all of the company’s data and context to allow for agentic workflows and agents to flourish.

[Tom Blomfield @t_blom]

Imagine replacing 90% of your employees with a team of geniuses who have no idea how your company operates.

Total chaos. Nothing works.

That’s what AI feels like today.

The missing piece is extracting all the domain knowledge from people’s heads and providing that as structured context to the models.

[dax @thdxr]

i have seen enough proof now that using a coding agent is a deep skill

it's confusing because the people you see heavily using them produce horrible results

but that's because it's a skill! you can get better and the ceiling seems pretty high - this is very exciting to me

[Anthony Diké @antdke]

I really haven't opened an IDE in about 6 months. It's wild.

Even when I want to read some code snippets while talking to Claude. I'll just ask it to output the snippets I need or I'll open a new terminal tab and just `cat` or `vim` that file.

No going back I think