Edmond Jhangiryan

curious mind.
building things.

[Edmond Jhangiryan @Edmondjhan]

new addiction

[shadcn @shadcn]

Every friend I talk to is overworked since AI. Working weekends. Always on their phone prompting. Kinda sad.

[Edmond Jhangiryan @Edmondjhan]

ran a full pentest on loomenia with @hakiraio

12 vulnerabilities. 5 critical. across rls, auth, and api layers. fixed and verified in 24h

paste a url, chat through findings in real-time. no scoping calls, no pdf reports, no back and forth

if you've been putting off security, give it a try. genuinely good experience

[Hakira ✨ @hakiraio]

Our AI agent completed an audit for @loomenia_ai.

Delivered a full-scope Web2 security audit, identifying vulnerabilities across application logic, authentication flows, APIs, and infrastructure, with actionable hardening recommendations.

[Edmond Jhangiryan @Edmondjhan]

here's what each one costs:

nano banana 2 (gemini 3.1 flash)
512px — $0.045
1K — $0.067
2K — $0.101
4K — $0.151

nano banana pro (gemini 3 pro)
1K/2K — $0.134
4K — $0.240

nano banana (gemini 2.5 flash)
1K — $0.039

nano banana 2 vs pro:
50% cheaper at 1K/2K
37% cheaper at 4K

nano banana 2 vs original:
72% more at 1K - but adds 2K, 4K, 14 reference images, search grounding, and thinking mode

[Edmond Jhangiryan @Edmondjhan]

google just dropped nano banana 2 (gemini 3.1 flash image)

same prompt, three models:
nano banana 2 → nano banana pro → nano banana

pro is the quality benchmark
2 is supposed to match it at half the cost and faster speed

here's how they actually compare ↓

[Edmond Jhangiryan @Edmondjhan]

google just dropped nano banana 2 (gemini 3.1 flash image)

same prompt, three models:
nano banana 2 → nano banana pro → nano banana

pro is the quality benchmark
2 is supposed to match it at half the cost and faster speed

here's how they actually compare ↓

[Google @Google]

Introducing Nano Banana 2: Our best image generation and editing model yet. 🍌

Pro-level quality, at Flash speed. Rolling out today across @GeminiApp, Search, and our developer and creativity tools.

[Edmond Jhangiryan @Edmondjhan]

tried arrow with a few use cases, logo generation with context about the project and template-based design with a reference image

for logos, gave it three style directions: clean & modern, symbolic, and playful

left is arrow, right is gemini
last image is a template with a reference

[QuiverAI @QuiverAI]

Today we're opening our public beta access to Arrow 1.0

A first of it's kind SVG AI model.
Turn your ideas into graphics.

[Edmond Jhangiryan @Edmondjhan]

other half of this, added an image library

every reference you uploaded is saved. click any slot, pick from history instead of re-uploading

works across generate, photoshoot, and templates

[Edmond Jhangiryan @Edmondjhan]

small ux cleanup, moved all reference images out of modals into a pinnable sidebar grid

pin what you use, hide what you don't

[Edmond Jhangiryan @Edmondjhan]

small ux cleanup, moved all reference images out of modals into a pinnable sidebar grid

pin what you use, hide what you don't

[Edmond Jhangiryan @Edmondjhan]

rebuilt entire pricing

had 3 plans that only differed by credit count.
the mid tier was a dead zone, just decision fatigue without real value.

rebuilt it as 2 plans with real feature gates.
each tier unlocks actual capabilities, not just "more of the same."

might add a third back later once usage patterns tell me something.
but right now, fewer options with clearer differences > more options.

[Edmond Jhangiryan @Edmondjhan]

https://t.co/Ve6lLrv5ya

[Edmond Jhangiryan @Edmondjhan]

one-click photoshoot

upload a reference image → get a full set of professional shots in minutes

same subject. different angles, poses, framing.

no studio. no photographer. no prompts.

just upload, generate, and post.

[Edmond Jhangiryan @Edmondjhan]

used to spend more time writing prompts than actually creating images.

most ai tools expect you to be a prompt engineer before you can be a creator.

here's what it looks like now:

1. upload a reference image
2. pick an environment
3. select a filter
4. hit generate

then conversationally make edits on the image - like "Turn into oil painting"

no prompts. just click and create.

[Edmond Jhangiryan @Edmondjhan]

i caught myself trying to “prompt correctly” instead of focusing on iterations

i’m building the ultimate UX where AI prompting doesn’t feel like work

a place to generate professional-level results, without sacrificing creative freedom

[Edmond Jhangiryan @Edmondjhan]

4% + 40¢ vs stripe's 2.9% + 30¢ sounds like more

but polar's fee already includes stripe's processing cost

the real MoR premium is about 1.1% + 10¢

when the alternative is incorporating, managing multi-country tax filings, and handling compliance yourself — it's not even close

important: this only works for digital products — SaaS subscriptions, digital downloads, license keys, courses. no physical goods, no services

if you're in one of stripe's 46 countries and selling digital products, the paperwork alone might be enough reason

if you're not? polar means you can actually get started

either way, spend that time building instead

[Edmond Jhangiryan @Edmondjhan]

building a digital product has never been easier

but getting paid for it across borders? still weirdly hard

here's why I chose polar sh over stripe — from someone who couldn't use stripe 🧵

[Edmond Jhangiryan @Edmondjhan]

I went with polar because stripe isn't available in my country

that was the starting point

but the more I looked into it, the more reasons showed up:

- no upfront cost to start
- 4% + 40¢ per transaction. for comparison, lemon squeezy charges 5% + 50¢
- simple onboarding. I created an org, answered a few questions, and was receiving money the same day
- open source

heads up: international cards add +1.5%, subscriptions add +0.5%. worth knowing upfront

[Edmond Jhangiryan @Edmondjhan]

curiosity + willingness to test what's new without clinging to comfort

the world moves fast, move with it

[Matt Shumer @mattshumer_]

https://t.co/ivXRKXJvQg

[Edmond Jhangiryan @Edmondjhan]

for anyone curious about the workflow
https://t.co/wdnJd7I8dG

[Edmond Jhangiryan @Edmondjhan]

was skeptical about claude's frontend-design skill

usually when you give ai rigid instructions it boxes them in. expected the same here

but after testing it against other tools and agents it's the best setup i've found for ui generation

went from generic ai components to this in a few iterations

here's what's working for me ↓

[Edmond Jhangiryan @Edmondjhan]

landing page is live

built the whole thing with claude's frontend-design skill

still refining copy, now wiring up payment and email

link in bio