i'm pooria. i build software that ships.
i build small software for people and agents, then write about what works.
four things i built and shipped
a small set of repeatable choices
one capability, several doors
the web interface is for people. the api, cli, and mcp tools give agents the same useful work.
one stack, reused on purpose
next.js on cloudflare, d1 for data, r2 for files, and queues for slow jobs. lessons move between products.
the person keeps the final say
replytosocial drafts and waits. the useful automation handles repetition without taking away judgment.
the work is only part of the week
showing up 
getting outside 
between builds
the ones i actually get
why build several small products?
each starts with one repeated problem. shared infrastructure lets me test a narrow idea without rebuilding everything.
why does each product have agent tools?
agents should not need to click through a screen built for people. useful work gets an api, cli, or mcp tool too.
can you look at what i am building?
yes. send one paragraph and a link through the contact page. no deck is needed for a first reply.
what shipped. what broke. what changed.
one email when there is something useful to say.