Anthony Lazkani

Building software. Chasing ideas

[Jihad @JihadNHammoud]

1/ We wanted to spend more time on the actual engineering side of a task:

understanding the problem, designing the solution, defining constraints and tests.

Then delegate as much of the implementation as possible to AI agents.

So we built a small coding harness for it.

[Anthony Lazkani @anthonylazkani2]

@jumainrink Even though ma bestaamil php bas thanks aal warning

[Jihad @JihadNHammoud]

If you receive a lot of emails daily, Demux will help you catch up 10x faster on your emails.

Reserve your early access now
👉🏻 https://t.co/UEZJC8bHWl

[Anthony Lazkani @anthonylazkani2]

@JihadNHammoud Let's goo !

[Jihad @JihadNHammoud]

You receive so many emails that it’s hard to keep track of everything.

We are building Demux to solve this.

Tell it what to track and how to see it. It builds a custom view that stays updated as new emails arrive.

Reserve early access: 50% off your first year.

https://t.co/xo2SxElajH

@anthonylazkani2

[Anthony Lazkani @anthonylazkani2]

I get why PHP is still widely used in legacy systems. But I was searching for Software Engineer jobs, I noticed startups created in 2026 still using PHP. It made me curious to be honest: why are teams choosing PHP for new projects? Maybe I am missing something.

[Anthony Lazkani @anthonylazkani2]

@jumainrink The best thing we have is family

[Anthony Lazkani @anthonylazkani2]

More people are finding Demux, but that doesn't mean we're getting closer to product-market fit.

Visitors are up 33%. Session duration is down 49%.

The numbers are a good reminder: getting people to visit is one problem. Getting them to care is another.

Back to learning.

[Jihad @JihadNHammoud]

I'm learning that "interesting" and "valuable" are very different words.

We launched https://t.co/xo2SxElajH and we're still figuring out whether people find Demux genuinely useful or just interesting enough to click.

PostHog is making that distinction painfully clear.

42% bounce rate.

People are landing. Some are curious. But curiosity isn't the goal.

The goal is: "I have this problem. I want this."

So now we're obsessing less over getting more traffic and more over understanding why people leave.

Building is easy compared to figuring out what people actually value.

[Jihad @JihadNHammoud]

@amasad Come to Paris. There’s a serious @Replit builder community growing here !

[Anthony Lazkani @anthonylazkani2]

@moveToMoonlight @JihadNHammoud Can you explain a bit more please?

[Jihad @JihadNHammoud]

Honest question for dev-founders: What was the hardest part about switching from "writing code" to "doing marketing," and what actually worked for you?

[Anthony Lazkani @anthonylazkani2]

Building the product is 10% of the work. What's the best piece of advice you have for devs trying to figure out the other 90% (distribution / marketing)?

[Jihad @JihadNHammoud]

@mattpocockuk's /grill-me skill is a good way to brainstorm architecture decisions.
Also pretty useful for learning if architecture isn’t something you do often.

https://t.co/WiLzJUGpHG

[Jihad @JihadNHammoud]

Great book if you’re getting started with building AI applications on top of foundation models.

https://t.co/AiL1LNNmXW

[Jihad @JihadNHammoud]

Try Genesis for designing agent workflows before implementation. Really useful for thinking through how to split the work between agents.

How are you structuring multi-agent workflows?

https://t.co/qN4LVKBTtH

[Jihad @JihadNHammoud]

The “harness matters more than the model” idea keeps coming up.

We spend so much time comparing models, while tools, context, memory, evals and the execution loop might be where most of the actual product advantage is built.

Definitely watching this one:
https://t.co/gkZibvKBHj

[Anthony Lazkani @anthonylazkani2]

@DanielSmidstrup Vibe coding for sure, at least you’re building something.

[Anthony Lazkani @anthonylazkani2]

Stop using LLM tokens for things 20 lines of TypeScript can solve.

Routing, JSON sanitation, and state management belong in deterministic code.

Save the model calls for actual ambiguous reasoning.

[Anthony Lazkani @anthonylazkani2]

@neeraj7105 Context limit