A little about me:
I’m Medeo, a solo founder and interaction designer in Rome.
By day I run @Jigoweb, my design studio, and build https://t.co/tsgG5QFlUw, an AI workspace for screenwriters (and more).
By night I make games and AI experiments.
Mostly a one-person operation, so I’m sharing it all here.
Say hi if you’re building too 👋
Human-Centered Design assumed a human was always at the controls.
Agents break that. The system can now click, buy, book, and reply without you in the loop.
So the job stops being "make the UI clear" and starts being "decide what the agent is allowed to do, when it must ask, and who owns the mistake."
That's not a visual problem. That's product politics wearing a UX badge.
Curious: are you still designing screens, or are you already designing permissions?
Agents don’t just “improve UX” they shift who holds power over actions and metrics. This piece argues that Human-Centered Design isn’t enough once software starts acting on our behalf, and that UX work has really become organizational strategy in disguise. https://t.co/p4YryZOG8j
New idea every week. Half-built stack. Zero cash.
Thesis is the bot that stops that.
One job: decide where capital goes so you focus on what can actually make money.
Stress-tests each bet, grades evidence, names the kill assumption, then GO / HOLD / KILL.
Import: https://t.co/8hpzXFQnUl
New update for my roguelike on @spawn 🦴
By popular request, there’s a new class: THE NECROMANCER 💀
Plus, tougher mobs across the board to crank up the difficulty.
Try the new class here: https://t.co/f2oITIkt6h
(Thanks Savi for the mini trailer)
Agents don’t just “improve UX” they shift who holds power over actions and metrics. This piece argues that Human-Centered Design isn’t enough once software starts acting on our behalf, and that UX work has really become organizational strategy in disguise. https://t.co/p4YryZOG8j
Designers: where does Human-Centered Design go in the era of AI agents?
Does it evolve into Human-Agent Centered Design, or does something else take its place?
Curious how you’re thinking about it.
Bookmark bait, not reality.
The instinct is right; “no catch” isn’t. Google buries mass‑generated pages, and I wouldn’t let a bot negotiate in my name. What actually works is a slow, niche directory on real data with a human OK on everything. Boring, but it compounds.
there is no catch and that scares me
Elon Musk posted an idea that hit 27 million views:
"stop building startups. buy boring internet real estate and let bots run it"
that's literally what Grok Bot does now. the whole setup costs $0 on top of the sub:
→ create a bot, name it Mogul
→ tell it to find one boring niche people search every day (it reads X and reddit complaints in real time, it's grok)
→ it builds a small directory site on its own computer and ships one useful page every night
→ it negotiates affiliate and referral deals itself, by email, in your name
→ publishing runs on a routine. telegram sends you the money reports
→ month one: traffic. month three: the first $217. it never stops after that
your laptop stays shut. the site grows anyway
every page is a tiny employee that works forever. it compounds like rent
some of them run 3 sites each and haven't opened a single one. they don't post about it, and now you know why
every month you wait, someone's bot ships another 30 pages into the niche you would've picked
full playbook below
bookmark this before someone takes your niche
Solid idea: shadow mode before letting it act is the right instinct. But in my setup, the bottleneck isn’t retry loops; it’s decisions waiting on me. And I wouldn’t hand the “ask the human” gate to a cheap external judge. The cheap version gets you 80% of it: a shared rule for all bots. Same failure twice, stop and report. Reuse what’s on disk first. Cap research at ~5 sources. No extra API needed. 🤷♂️