Buddy

Engineering manager | 8+ years in software
Writing about AI, code, teams, and career leverage
Still prefers code to status meetings

[Buddy @dyorbuddy]

https://t.co/PX6yCF2wxa

[Buddy @dyorbuddy]

The funny thing about AI agents:

the prototype can look useful in two days.

Then you add deadlines, token budgets, roles, fallback models, logs, caching, and loop protection.

Suddenly it stops looking like a demo and starts looking like software engineering.

[Buddy @dyorbuddy]

https://t.co/PX6yCF2wxa

[Buddy @dyorbuddy]

@og_miracles Exactly.

A demo can ignore cost.

A real workflow has deadlines, retries, stale context, and a token bill.

That is where the agent stops looking cute.

[Buddy @dyorbuddy]

Everyone wants to build AI agents until the agent spends 25 minutes reading docs, burns tokens, hits 503, repeats the same search, and still gives a weak answer.

The useful part starts after the cute demo.

I wrote down what actually makes an internal AI agent work:

[Buddy @dyorbuddy]

https://t.co/PX6yCF2wxa

[Buddy @dyorbuddy]

@JustJerry121 Yeah, context plumbing is the part people underestimate.

Once agents touch real folders, the hard question becomes simple: what should they see right now?

Canvas + file graph sounds like a solid direction.

[Buddy @dyorbuddy]

https://t.co/PX6yCF2wxa

[Buddy @dyorbuddy]

@thegreatest_sv Sure, it worth it 100%

[Buddy @dyorbuddy]

@iamigorekk AI tokens are smth like a future money

[Buddy @dyorbuddy]

@claudeai @BenJames_____ Good insight from Ben. Everyone can open the world to themselves with tools like Claude.

[Buddy @dyorbuddy]

A typical vibe coder day was basically already shown in "Fired on Mars".

Yes, Jeff is a designer, but honestly, close enough.

Great show btw.

[Buddy @dyorbuddy]

@FabrizioRomano @MenInBlazers Gonna listen to it. Not sure Mourinho will stay more than 2 seasons in Madrid

[Buddy @dyorbuddy]

This article is useful because it explains the boring reason why people burn Claude Max so fast.

Most users treat Claude like one endless chat for everything.

Small question about a function.
Same chat.

Code review.
Same chat.

Refactor.
Same chat.

Marketing draft.
Same chat.

After 30-40 messages, the model is reading a huge pile of old context before answering a tiny new question. Add files, tool results, terminal output, and a few wrong turns. Now every prompt becomes expensive.

The second problem is effort level.

People buy Max and leave everything on the strongest setting because it “feels safer”.

Simple formatting? Max.
Tiny bug? Max.
One-line question? Max.

That is how you spend architecture-level reasoning on tasks that needed a quick answer.

The setup that makes more sense:

low effort for small questions and formatting
medium or high for daily coding
max only for deep debugging, architecture, or ugly multi-file changes
compact long sessions
split work by feature
ask for a plan before large edits
keep project rules in CLAUDE.md so you don’t explain the stack every day

This is the part many people miss with AI coding tools.

Buying a bigger plan helps, but bad workflow will still burn it fast.

The video below adds the practical side: 18 Claude Code token hacks for smaller sessions, cleaner context, and fewer wasted prompts.

[kaize @0x_kaize]

https://t.co/c94ruVOMFA

[Buddy @dyorbuddy]

Every model release raises the tide.

The money is in boats for boring, expensive workflows: law, finance, compliance, insurance.

Legal AI works because the customer already pays for research, drafts, review, and risk.

AI just compresses the loop.

[Claude @claudeai]

Interpreting law is one of the oldest jobs in the world. @MaxJunestrand, co-founder and CEO of @WeAreLegora, is bringing it into its next era with Claude.

His bet: every new model release raises the tide, and Legora is building the boats for everyone else.

[Buddy @dyorbuddy]

@claudeai @MaxJunestrand @WeAreLegora Legal AI gets interesting when it stops being a chatbot and starts fitting into the actual work: research, citations, drafts, review, risk checks.

That is where lawyers will pay.

[Buddy @dyorbuddy]

This looks stupid until you realize what is happening.

This 21-old guy is running multiple Claude Code agents like a small dev team.

One agent codes.
One writes tests.
One reviews.
One updates docs.
One prepares deployment.

Not “AI autocomplete”.

More like parallel engineering work.

I broke down the setup here 👇

[Buddy @dyorbuddy]

https://t.co/emoBmnfexl

[Buddy @dyorbuddy]

@claudeai @pirroh @Replit Natural language makes the first version easier.

The hard part is still the same: knowing what to build, what to remove, and why users would care.

[Buddy @dyorbuddy]

@elonmusk Earth-only is a single point of failure.
The Moon is a fast backup, but still close to the same blast radius.
Mars is painful and slow, but it changes the failure domain.

Basically fault-tolerant architecture, but for civilization.

[Buddy @dyorbuddy]

@bonsaixbt Going to read that article and watch video as well now

Thanks for your work

[Buddy @dyorbuddy]

@bonsaixbt For sure you'll find. Thanks for your attention

[Buddy @dyorbuddy]

An Anthropic engineer showed what “agent teams” actually look like in Claude Code.

Not theory.

One person running agents for code, tests, review, docs, and deployment in parallel.

The interesting part is not the demo itself, but the operating system behind it:

> when one agent is enough
> when you need a full team
> how to split work so agents don’t block each other
> how to decide what each agent owns

This is the part most people miss.

So I wrote a practical guide on building Claude Code agent teams that actually work.

Full guide below 👇

[Buddy @dyorbuddy]

https://t.co/emoBmnfexl

[Buddy @dyorbuddy]

@claudeai We automated the easy part and somehow kept the whiteboard interview part.