Roman Imankulov

20 years of Python and still debugging with print(). Building Smello to fix that

[Roman Imankulov @rdotpy]

The Claude Agent SDK lets you flip the execution model: the agent drives the logic, Python handles hard logic parts.

I rewrote my Claude Code news reader with the Agent SDK. Everything lives in a single file now, and it feels like a normal Python project again.

The code still behaves like a Claude Code session, but it's easier now to deploy and productize.

https://t.co/n684pgflsd
(follow-up to https://t.co/pDh4fBpVIQ)

I have a draft doing the same thing with Pi, and Deep Agents from LangChain is next.

[Roman Imankulov @rdotpy]

@scukko Thanks! Yeah, the example is intentionally bare-bones, but the real system can definitely go well beyond that

[Roman Imankulov @rdotpy]

I've been experimenting with inverting how I use LLM. In my recent app, the agent loop drives the logic, and code only shows up where I need it.

I documented how it works. Feedback so far has been one "dude, just use OpenClaw" and "dude, just use Pi."

https://t.co/qZvhEkfi9z

[Roman Imankulov @rdotpy]

@philkellr Considered https://t.co/SaTaw3epn5?

[Roman Imankulov @rdotpy]

@theroborourke I guess, to take full advantage of it, you need to live in Poland, right? Or you’ll eventually pay the rest as the income tax of your country of residence.

[Roman Imankulov @rdotpy]

@dqmonn It feels a lot like Toptal. There's a matcher that reaches out to talent, filters the results, and gets back to you with a few options.

[Roman Imankulov @rdotpy]

@robwalling A rewrite is often disguised as "let's replace this foundational part of the system (ORM, framework, you name it) with an equivalent," leading to similar consequences.

[Roman Imankulov @rdotpy]

@divine_economy Isn't that the same strategy Stripe used to become wildly successful?

[Roman Imankulov @rdotpy]

@dqmonn When I worked at a hosting company, we had a highly successful referral program. Web studios would bring their clients to our hosting and receive a share of what the customers paid.

[Roman Imankulov @rdotpy]

Promised a quick and dirty solution. Delivered on dirty. Failed on quick.

[Roman Imankulov @rdotpy]

My customers asked for a faster horse. Instead, I created them a GPT interface app.

[Roman Imankulov @rdotpy]

Monday procrastination trick — when creating @NotionHQ page, spend more time choosing emoji than writing content

[Roman Imankulov @rdotpy]

If it hurts, do it more often.

How we introduced feature flags at @BuildingRadar and what have we learned so far. Special thanks to @DevCycleHQ, the feature management provider of our choice.

https://t.co/0pais60Pj1

[Roman Imankulov @rdotpy]

That fear of a connection timeout on the last step of a purchase while the browser is still loading the page.

[Roman Imankulov @rdotpy]

Following the most exciting opportunity works well if you are an expert in a regular environment. Otherwise, you enthusiastically follow a random choice. Thinking aloud why HELL YEAH strategy may not be the best move and what to do about it.

https://t.co/ytcJr4FQuP

[Roman Imankulov @rdotpy]

@brennakL My solution

[Roman Imankulov @rdotpy]

@rochacbruno Beautiful. Also, I like their "why?" part that concisely outlines alternatives and the tech stack behind the tool.

[Roman Imankulov @rdotpy]

@amix3k Just a few days ago, we discussed with Evert if GPT-3 could help answer support questions and found that it probably can't because its answers to Todoist-specific questions are plain wrong. With pre-training, this can be a whole new story.

[Roman Imankulov @rdotpy]

@amix3k Oh, nice! I always thought of GPT-3 as a black box whose output depends only on the prompt.

[Roman Imankulov @rdotpy]

@zeeg Really happy to see this improvement! I've yet to see a non-playground Sentry installation not cluttered with timeout or connection errors that team can't resolve but afraid to ignore.

Among these, auto-ignoring is probably the biggest change, and I welcome it.