Stavros Thalassinos - building Akova

Building Akova — private, on-device dictation for Mac. Your voice never leaves your machine. No cloud, no subscription. Solo dev, 🇬🇷. Building in public.

[Stavros Thalassinos - building Akova @Stavthal]

Powerful and timely vision, Demis. The comparison to fire and electricity hits hard. AGI could truly unlock abundance if we get the guardrails right. Love the concrete proposal for a US-led Frontier AI Standards Body with independent evals, pre-release reviews, and the ability to slow things down when needed. Cautious optimism in action. This feels like the responsible path to steer toward that golden age of discovery. Excited (and a bit nervous) to see how the community and policymakers respond. 👏

[Stavros Thalassinos - building Akova @Stavthal]

I'm a Greek software engineer building this solo, in public — my first real attempt at something
of my own. If "private, buy-once Mac tools" is your thing, follow along. I'll be shipping and sharing
the whole way. Waitlist opening soon.

[Stavros Thalassinos - building Akova @Stavthal]

Day 1 of building in public. I'm making Akova — a Mac dictation app that runs 100% on your
device. No cloud, no subscription. Your voice never leaves your Mac. Here's why 🧵

[Stavros Thalassinos - building Akova @Stavthal]

So that's Akova: press a key, talk, clean text lands in whatever app you're in — transcribed
on-device, cleaned up on-device, pasted at your cursor. Offline. Private. Yours.

[Stavros Thalassinos - building Akova @Stavthal]

@cuteonlyvibes Silver Gin. Has a nice ring to it.

[Stavros Thalassinos - building Akova @Stavthal]

@NVIDIAGeForce GeForce Season

[Stavros Thalassinos - building Akova @Stavthal]

@LancesEmporium well as a kind of casual player, I am already using maximum blue rarity thingies, as I am playing solo mostly. I have already completed all the quests so its just drop in, loot, kill/die, rinse and repeat. Resets help that part too imho.

[Stavros Thalassinos - building Akova @Stavthal]

@GamingMerchant_ @grok how do you farm blueprints in arc raiders

[Stavros Thalassinos - building Akova @Stavthal]

@krishdotdev AI can steal your job only if you are dumber than it.

[Stavros Thalassinos - building Akova @Stavthal]

@webdesignerng As a web dev that really likes python as it was the first language I ever wrote, I have to say bye bye python.

[Stavros Thalassinos - building Akova @Stavthal]

@_devJNS Learned python at school, JavaScript got me in the industry as it gave me my first job, still using it today (although I have mostly transitioned to typescript).

[Stavros Thalassinos - building Akova @Stavthal]

@voidmode @NoahEpstein_ No replies. Guess why.

[Stavros Thalassinos - building Akova @Stavthal]

@opsided @NoahEpstein_ Finally someone who is not blinded by the AI bubble.

[AIOPS | @opsided]

This is the same LinkedIn rage-bait format from earlier, just reskinned for Cursor.

What’s actually true:

Cursor Composer is genuinely impressive. Multi-agent coding with browser testing is a real capability leap. The speed improvements are real for certain workflows.

What’s bullshit:

“Replace 3 junior devs ($450K/year)”, Junior devs don’t make $150K each, and if your juniors are that expensive you’re already overpaying or in SF/NYC where $240/year of tooling wasn’t your constraint anyway.

“8 agents running parallel in 30 seconds”, for what? Scaffolding a CRUD app? Sure. Building anything with complex state, integrations, or edge cases? No chance. This is demo magic, not production reality.

“Test their own code in a native browser”, cool feature. But automated testing has existed forever. The hard part isn’t running tests, it’s knowing what to test and interpreting failures.

“99.9% cost reduction for better output”, If the output was actually better, those companies would have zero devs. They don’t, because AI still can’t handle ambiguity, changing requirements, or the 1000 tiny decisions that separate a prototype from a product.

The grift signals:

“Comment ‘COMPOSER’ and I’ll send…” -Lead gen funnel
“Your competition is still hiring. Time to bury them.” -Fear-based urgency
Absurd numbers presented without context

The real story is:

Cursor is legitimately good and getting better. It makes individual developers more productive. Small teams can ship faster. But it’s not replacing teams, it’s changing what those teams focus on.

The people getting the most value from Cursor are experienced devs who know what to build and can validate the output. Not people who think it’s a dev team replacement.

[trash @trashh_dev]

oncall engineers:

i can’t help you. it’s an aws problem

[Stavros Thalassinos - building Akova @Stavthal]

@mark_k Does it run Doom though