writing.
everything i think in public — long essays and short posts, one feed. agents, founding, and the road to an agency that runs itself.
agency owners do not need more ai tools. they need an ai workspace
ai tools for agency owners is not just a search term for aios. it is the operating pattern firaz is building around agents, loops, terminals, and real work.
read →how i use claude code, codex, and aios in one workspace
claude code and codex are strong, but the real unlock is the workspace around them. this is how i run agents, terminals, browser panes, files, and notes inside aios.
read →i'm building an agency that runs without people. here's why.
every agency sells the same thing — humans renting their hours. i think that whole model is about to break. this is the thesis behind aios.
read →i got 0 to 30 paying users in 2 hours. here's exactly what happened.
a friend called on a tuesday wanting a rebrand. by the next morning the thing had 30 paying users. this is the unglamorous playbook — no audience, no ads, no growth hack.
read →first time in my life a product i made went from 0 → 30 paying users in 2 hours watched the number climb in real time — 0 → 800 → 1000 → 1500. while i was live debugging and onboarding people as they came in
backstory: tuesday a friend called out of nowhere. wanted to rebrand a trading bot i'd built for him months ago — primusgpt, my quiet "sleeper" build from september that night i rebuilt the whole thing.
bot, landing page, payments. one night and here's what actually hit me every other thing i've built, i had to be the whole machine. find the market, build it, handle distribution, close the sale. all me this time i just built. my partner owns the distribution. that's the entire difference
i don't even own most of it only 10% but i built it, got paid and kept a small piece but it taught me the thing i kept getting wrong: i don't have to do everything make a good product. let someone great at distribution carry it
i replaced a $6k/month agency with ai for $200. here's the exact 30-day migration.
the cost math is brutal once you see it — a full-service retainer is ~$72k a year, the ai stack that does the execution is ~$2.4k. here's how the swap actually goes, week by week, and where it bites.
read →can ai actually replace a marketing agency? i run one on agents — here's where it breaks.
the loudest objection online is 'the people saying agents replace marketers have never marketed anything — prove it.' fair. so here's the honest version from someone who did, including the parts that don't work.
read →stop shipping features. ship agents.
a feature waits for you to use it. an agent does the work while you sleep. the difference compounds faster than anyone expects.
read →the ai automation in my business that actually holds up at scale
everyone shows you the demo. nobody shows you what survives once you're running a lot of it. here's the automation in my company that didn't fall apart at volume — and the kind that always does.
read →me: make me a story for my product launch please aios:
launching aios on product hunt today. i built it because my ai workflow was getting messy: claude in one window codex in another terminals everywhere browser tabs for docs notes and files somewhere else the agents are powerful. the workspace is broken. aios is my attempt at the missing shell: a native desktop app for agentic work. open source. mac build live. windows port in progress. would mean a lot if you support
demo + download: https://aios.adleticagency.com github: https://github.com/ferazfhansurie/aios-superapp
i'm not building an agency. i'm building the operating system agencies run on.
the agency is the demo. the real product is aios — the operating system my company runs on, and the one i want every other agency to run on too.
read →stop treating ai like a magic trick. treat it like software.
the reason most ai agents become unmaintainable is that people build them as clever prompts instead of as software. the fix isn't a better model — it's engineering discipline you already know.
read →the best way to actually learn ai agents in 2026 (from someone shipping them daily)
the honest path isn't a course or a framework tier-list. it's building one ugly agent that does one real thing, in production, this week. here's the route i'd take if i started over.
read →marketing agencies don't need more dashboards. they need fewer humans in the loop.
every agency tool promises another dashboard. but the owner doesn't want to watch numbers move — they want the work done. that gap is the whole opportunity.
read →marketing agencies do not need more dashboards. the owner becomes the routing layer when leads, client updates, campaign blockers, reports, and content plans all live in different tools. i'm building aios around that loop now: lead inbox -> client command center -> reporting agent.
i stopped chasing investors and took a job at a cooking-oil company.
the canon event nobody posts about. i quit the investor hunt, took a fullstack job at a used-cooking-oil company, and it was the best thing that happened to the agency.
read →post major canon event update and things i am working on: no more investor had to get a job now working at a used cooking oil collection company as a fullstack dev faeez and jullien left but honestly means more money for me but also more work which is good ofcourse some of my clients churn but its for the better can focus on clients that really matter and the job